Showing 1–50 of 124 results for author: White, M

Position: Benchmarking is Limited in Reinforcement Learning Research

Authors: Scott M. Jordan, Adam White, Bruno Castro da Silva, Martha White, Philip S. Thomas

Abstract: Novel reinforcement learning algorithms, or improvements on existing ones, are commonly justified by evaluating their performance on benchmark environments and are compared to an ever-changing set of standard algorithms. However, despite numerous calls for improvements, experimental practices continue to produce misleading or unsupported claims. One reason for the ongoing substandard practices is… ▽ More Novel reinforcement learning algorithms, or improvements on existing ones, are commonly justified by evaluating their performance on benchmark environments and are compared to an ever-changing set of standard algorithms. However, despite numerous calls for improvements, experimental practices continue to produce misleading or unsupported claims. One reason for the ongoing substandard practices is that conducting rigorous benchmarking experiments requires substantial computational time. This work investigates the sources of increased computation costs in rigorous experiment designs. We show that conducting rigorous performance benchmarks will likely have computational costs that are often prohibitive. As a result, we argue for using an additional experimentation paradigm to overcome the limitations of benchmarking. △ Less

Submitted 23 June, 2024; originally announced June 2024.

Comments: 19 pages, 13 figures, The Forty-first International Conference on Machine Learning (ICML 2024)

Demystifying the Recency Heuristic in Temporal-Difference Learning

Authors: Brett Daley, Marlos C. Machado, Martha White

Abstract: The recency heuristic in reinforcement learning is the assumption that stimuli that occurred closer in time to an acquired reward should be more heavily reinforced. The recency heuristic is one of the key assumptions made by TD($λ$), which reinforces recent experiences according to an exponentially decaying weighting. In fact, all other widely used return estimators for TD learning, such as $n$-st… ▽ More The recency heuristic in reinforcement learning is the assumption that stimuli that occurred closer in time to an acquired reward should be more heavily reinforced. The recency heuristic is one of the key assumptions made by TD($λ$), which reinforces recent experiences according to an exponentially decaying weighting. In fact, all other widely used return estimators for TD learning, such as $n$-step returns, satisfy a weaker (i.e., non-monotonic) recency heuristic. Why is the recency heuristic effective for temporal credit assignment? What happens when credit is assigned in a way that violates this heuristic? In this paper, we analyze the specific mathematical implications of adopting the recency heuristic in TD learning. We prove that any return estimator satisfying this heuristic: 1) is guaranteed to converge to the correct value function, 2) has a relatively fast contraction rate, and 3) has a long window of effective credit assignment, yet bounded worst-case variance. We also give a counterexample where on-policy, tabular TD methods violating the recency heuristic diverge. Our results offer some of the first theoretical evidence that credit assignment based on the recency heuristic facilitates learning. △ Less

Submitted 18 June, 2024; originally announced June 2024.

Comments: RLC 2024. 18 pages, 8 figures, 1 table

A New View on Planning in Online Reinforcement Learning

Authors: Kevin Roice, Parham Mohammad Panahi, Scott M. Jordan, Adam White, Martha White

Abstract: This paper investigates a new approach to model-based reinforcement learning using background planning: mixing (approximate) dynamic programming updates and model-free updates, similar to the Dyna architecture. Background planning with learned models is often worse than model-free alternatives, such as Double DQN, even though the former uses significantly more memory and computation. The fundament… ▽ More This paper investigates a new approach to model-based reinforcement learning using background planning: mixing (approximate) dynamic programming updates and model-free updates, similar to the Dyna architecture. Background planning with learned models is often worse than model-free alternatives, such as Double DQN, even though the former uses significantly more memory and computation. The fundamental problem is that learned models can be inaccurate and often generate invalid states, especially when iterated many steps. In this paper, we avoid this limitation by constraining background planning to a set of (abstract) subgoals and learning only local, subgoal-conditioned models. This goal-space planning (GSP) approach is more computationally efficient, naturally incorporates temporal abstraction for faster long-horizon planning and avoids learning the transition dynamics entirely. We show that our GSP algorithm can propagate value from an abstract space in a manner that helps a variety of base learners learn significantly faster in different domains. △ Less

Submitted 3 June, 2024; originally announced June 2024.

Comments: Published in the Planning and Reinforcement Learning Workshop at ICAPS 2024. arXiv admin note: text overlap with arXiv:2206.02902

Granite Code Models: A Family of Open Foundation Models for Code Intelligence

Authors: Mayank Mishra, Matt Stallone, Gaoyuan Zhang, Yikang Shen, Aditya Prasad, Adriana Meza Soria, Michele Merler, Parameswaran Selvam, Saptha Surendran, Shivdeep Singh, Manish Sethi, Xuan-Hong Dang, Pengyuan Li, Kun-Lung Wu, Syed Zawad, Andrew Coleman, Matthew White, Mark Lewis, Raju Pavuluri, Yan Koyfman, Boris Lublinsky, Maximilien de Bayser, Ibrahim Abdelaziz, Kinjal Basu, Mayank Agarwal , et al. (21 additional authors not shown)

Abstract: Large Language Models (LLMs) trained on code are revolutionizing the software development process. Increasingly, code LLMs are being integrated into software development environments to improve the productivity of human programmers, and LLM-based agents are beginning to show promise for handling complex tasks autonomously. Realizing the full potential of code LLMs requires a wide range of capabili… ▽ More Large Language Models (LLMs) trained on code are revolutionizing the software development process. Increasingly, code LLMs are being integrated into software development environments to improve the productivity of human programmers, and LLM-based agents are beginning to show promise for handling complex tasks autonomously. Realizing the full potential of code LLMs requires a wide range of capabilities, including code generation, fixing bugs, explaining and documenting code, maintaining repositories, and more. In this work, we introduce the Granite series of decoder-only code models for code generative tasks, trained with code written in 116 programming languages. The Granite Code models family consists of models ranging in size from 3 to 34 billion parameters, suitable for applications ranging from complex application modernization tasks to on-device memory-constrained use cases. Evaluation on a comprehensive set of tasks demonstrates that Granite Code models consistently reaches state-of-the-art performance among available open-source code LLMs. The Granite Code model family was optimized for enterprise software development workflows and performs well across a range of coding tasks (e.g. code generation, fixing and explanation), making it a versatile all around code model. We release all our Granite Code models under an Apache 2.0 license for both research and commercial use. △ Less

Submitted 7 May, 2024; originally announced May 2024.

Comments: Corresponding Authors: Rameswar Panda, Ruchir Puri; Equal Contributors: Mayank Mishra, Matt Stallone, Gaoyuan Zhang

K-percent Evaluation for Lifelong RL

Authors: Golnaz Mesbahi, Parham Mohammad Panahi, Olya Mastikhina, Martha White, Adam White

Abstract: In continual or lifelong reinforcement learning, access to the environment should be limited. If we aspire to design algorithms that can run for long periods, continually adapting to new, unexpected situations, then we must be willing to deploy our agents without tuning their hyperparameters over the agent's entire lifetime. The standard practice in deep RL, and even continual RL, is to assume unf… ▽ More In continual or lifelong reinforcement learning, access to the environment should be limited. If we aspire to design algorithms that can run for long periods, continually adapting to new, unexpected situations, then we must be willing to deploy our agents without tuning their hyperparameters over the agent's entire lifetime. The standard practice in deep RL, and even continual RL, is to assume unfettered access to the deployment environment for the full lifetime of the agent. In this paper, we propose a new approach for evaluating lifelong RL agents where only k percent of the experiment data can be used for hyperparameter tuning. We then conduct an empirical study of DQN and SAC across a variety of continuing and non-stationary domains. We find agents generally perform poorly when restricted to k-percent tuning, whereas several algorithmic mitigations designed to maintain network plasticity perform surprisingly well. △ Less

Submitted 25 May, 2024; v1 submitted 2 April, 2024; originally announced April 2024.

The Model Openness Framework: Promoting Completeness and Openness for Reproducibility, Transparency, and Usability in Artificial Intelligence

Authors: Matt White, Ibrahim Haddad, Cailean Osborne, Xiao-Yang Liu Yanglet, Ahmed Abdelmonsef, Sachin Varghese

Abstract: Generative AI (GAI) offers unprecedented opportunities for research and innovation, but its commercialization has raised concerns about transparency, reproducibility, and safety. Many open GAI models lack the necessary components for full understanding and reproducibility, and some use restrictive licenses whilst claiming to be ``open-source''. To address these concerns, we propose the Model Openn… ▽ More Generative AI (GAI) offers unprecedented opportunities for research and innovation, but its commercialization has raised concerns about transparency, reproducibility, and safety. Many open GAI models lack the necessary components for full understanding and reproducibility, and some use restrictive licenses whilst claiming to be ``open-source''. To address these concerns, we propose the Model Openness Framework (MOF), a ranked classification system that rates machine learning models based on their completeness and openness, following principles of open science, open source, open data, and open access. The MOF requires specific components of the model development lifecycle to be included and released under appropriate open licenses. This framework aims to prevent misrepresentation of models claiming to be open, guide researchers and developers in providing all model components under permissive licenses, and help individuals and organizations identify models that can be safely adopted without restrictions. By promoting transparency and reproducibility, the MOF combats ``openwashing'' practices and establishes completeness and openness as primary criteria alongside the core tenets of responsible AI. Wide adoption of the MOF will foster a more open AI ecosystem, benefiting research, innovation, and adoption of state-of-the-art models. △ Less

Submitted 3 June, 2024; v1 submitted 20 March, 2024; originally announced March 2024.

Comments: 22 pages

Investigating the Histogram Loss in Regression

Authors: Ehsan Imani, Kai Luedemann, Sam Scholnick-Hughes, Esraa Elelimy, Martha White

Abstract: It is becoming increasingly common in regression to train neural networks that model the entire distribution even if only the mean is required for prediction. This additional modeling often comes with performance gain and the reasons behind the improvement are not fully known. This paper investigates a recent approach to regression, the Histogram Loss, which involves learning the conditional distr… ▽ More It is becoming increasingly common in regression to train neural networks that model the entire distribution even if only the mean is required for prediction. This additional modeling often comes with performance gain and the reasons behind the improvement are not fully known. This paper investigates a recent approach to regression, the Histogram Loss, which involves learning the conditional distribution of the target variable by minimizing the cross-entropy between a target distribution and a flexible histogram prediction. We design theoretical and empirical analyses to determine why and when this performance gain appears, and how different components of the loss contribute to it. Our results suggest that the benefits of learning distributions in this setup come from improvements in optimization rather than learning a better representation. We then demonstrate the viability of the Histogram Loss in common deep learning applications without a need for costly hyperparameter tuning. △ Less

Submitted 20 February, 2024; originally announced February 2024.

Comments: 50 pages

When is Tree Search Useful for LLM Planning? It Depends on the Discriminator

Authors: Ziru Chen, Michael White, Raymond Mooney, Ali Payani, Yu Su, Huan Sun

Abstract: In this paper, we examine how large language models (LLMs) solve multi-step problems under a language agent framework with three components: a generator, a discriminator, and a planning method. We investigate the practical utility of two advanced planning methods, iterative correction and tree search. We present a comprehensive analysis of how discrimination accuracy affects the overall performanc… ▽ More In this paper, we examine how large language models (LLMs) solve multi-step problems under a language agent framework with three components: a generator, a discriminator, and a planning method. We investigate the practical utility of two advanced planning methods, iterative correction and tree search. We present a comprehensive analysis of how discrimination accuracy affects the overall performance of agents when using these two methods or a simpler method, re-ranking. Experiments on two tasks, text-to-SQL parsing and mathematical reasoning, show that: (1) advanced planning methods demand discriminators with at least 90% accuracy to achieve significant improvements over re-ranking; (2) current LLMs' discrimination abilities have not met the needs of advanced planning methods to achieve such improvements; (3) with LLM-based discriminators, advanced planning methods may not adequately balance accuracy and efficiency. For example, compared to the other two methods, tree search is at least 10--20 times slower but leads to negligible performance gains, which hinders its real-world applications. Code and data are available at https://github.com/OSU-NLP-Group/llm-planning-eval. △ Less

Submitted 6 June, 2024; v1 submitted 16 February, 2024; originally announced February 2024.

Comments: ACL 2024 main

What to Do When Your Discrete Optimization Is the Size of a Neural Network?

Authors: Hugo Silva, Martha White

Abstract: Oftentimes, machine learning applications using neural networks involve solving discrete optimization problems, such as in pruning, parameter-isolation-based continual learning and training of binary networks. Still, these discrete problems are combinatorial in nature and are also not amenable to gradient-based optimization. Additionally, classical approaches used in discrete settings do not scale… ▽ More Oftentimes, machine learning applications using neural networks involve solving discrete optimization problems, such as in pruning, parameter-isolation-based continual learning and training of binary networks. Still, these discrete problems are combinatorial in nature and are also not amenable to gradient-based optimization. Additionally, classical approaches used in discrete settings do not scale well to large neural networks, forcing scientists and empiricists to rely on alternative methods. Among these, two main distinct sources of top-down information can be used to lead the model to good solutions: (1) extrapolating gradient information from points outside of the solution set (2) comparing evaluations between members of a subset of the valid solutions. We take continuation path (CP) methods to represent using purely the former and Monte Carlo (MC) methods to represent the latter, while also noting that some hybrid methods combine the two. The main goal of this work is to compare both approaches. For that purpose, we first overview the two classes while also discussing some of their drawbacks analytically. Then, on the experimental section, we compare their performance, starting with smaller microworld experiments, which allow more fine-grained control of problem variables, and gradually moving towards larger problems, including neural network regression and neural network pruning for image classification, where we additionally compare against magnitude-based pruning. △ Less

Submitted 15 February, 2024; originally announced February 2024.

Comments: Submitted to JMLR

Averaging $n$-step Returns Reduces Variance in Reinforcement Learning

Authors: Brett Daley, Martha White, Marlos C. Machado

Abstract: Multistep returns, such as $n$-step returns and $λ$-returns, are commonly used to improve the sample efficiency of reinforcement learning (RL) methods. The variance of the multistep returns becomes the limiting factor in their length; looking too far into the future increases variance and reverses the benefits of multistep learning. In our work, we demonstrate the ability of compound returns -- we… ▽ More Multistep returns, such as $n$-step returns and $λ$-returns, are commonly used to improve the sample efficiency of reinforcement learning (RL) methods. The variance of the multistep returns becomes the limiting factor in their length; looking too far into the future increases variance and reverses the benefits of multistep learning. In our work, we demonstrate the ability of compound returns -- weighted averages of $n$-step returns -- to reduce variance. We prove for the first time that any compound return with the same contraction modulus as a given $n$-step return has strictly lower variance. We additionally prove that this variance-reduction property improves the finite-sample complexity of temporal-difference learning under linear function approximation. Because general compound returns can be expensive to implement, we introduce two-bootstrap returns which reduce variance while remaining efficient, even when using minibatched experience replay. We conduct experiments showing that compound returns often increase the sample efficiency of $n$-step deep RL agents like DQN and PPO. △ Less

Submitted 5 June, 2024; v1 submitted 6 February, 2024; originally announced February 2024.

Comments: ICML 2024. 27 pages, 7 figures, 3 tables

Differentially Private Low-Rank Adaptation of Large Language Model Using Federated Learning

Authors: Xiao-Yang Liu, Rongyi Zhu, Daochen Zha, Jiechao Gao, Shan Zhong, Matt White, Meikang Qiu

Abstract: The surge in interest and application of large language models (LLMs) has sparked a drive to fine-tune these models to suit specific applications, such as finance and medical science. However, concerns regarding data privacy have emerged, especially when multiple stakeholders aim to collaboratively enhance LLMs using sensitive data. In this scenario, federated learning becomes a natural choice, al… ▽ More The surge in interest and application of large language models (LLMs) has sparked a drive to fine-tune these models to suit specific applications, such as finance and medical science. However, concerns regarding data privacy have emerged, especially when multiple stakeholders aim to collaboratively enhance LLMs using sensitive data. In this scenario, federated learning becomes a natural choice, allowing decentralized fine-tuning without exposing raw data to central servers. Motivated by this, we investigate how data privacy can be ensured in LLM fine-tuning through practical federated learning approaches, enabling secure contributions from multiple parties to enhance LLMs. Yet, challenges arise: 1) despite avoiding raw data exposure, there is a risk of inferring sensitive information from model outputs, and 2) federated learning for LLMs incurs notable communication overhead. To address these challenges, this article introduces DP-LoRA, a novel federated learning algorithm tailored for LLMs. DP-LoRA preserves data privacy by employing a Gaussian mechanism that adds noise in weight updates, maintaining individual data privacy while facilitating collaborative model training. Moreover, DP-LoRA optimizes communication efficiency via low-rank adaptation, minimizing the transmission of updated weights during distributed training. The experimental results across medical, financial, and general datasets using various LLMs demonstrate that DP-LoRA effectively ensures strict privacy constraints while minimizing communication overhead. △ Less

Submitted 2 June, 2024; v1 submitted 29 December, 2023; originally announced December 2023.

Comments: 21 pages, 1 figure, 19 tables

When is Offline Policy Selection Sample Efficient for Reinforcement Learning?

Authors: Vincent Liu, Prabhat Nagarajan, Andrew Patterson, Martha White

Abstract: Offline reinforcement learning algorithms often require careful hyperparameter tuning. Consequently, before deployment, we need to select amongst a set of candidate policies. As yet, however, there is little understanding about the fundamental limits of this offline policy selection (OPS) problem. In this work we aim to provide clarity on when sample efficient OPS is possible, primarily by connect… ▽ More Offline reinforcement learning algorithms often require careful hyperparameter tuning. Consequently, before deployment, we need to select amongst a set of candidate policies. As yet, however, there is little understanding about the fundamental limits of this offline policy selection (OPS) problem. In this work we aim to provide clarity on when sample efficient OPS is possible, primarily by connecting OPS to off-policy policy evaluation (OPE) and Bellman error (BE) estimation. We first show a hardness result, that in the worst case, OPS is just as hard as OPE, by proving a reduction of OPE to OPS. As a result, no OPS method can be more sample efficient than OPE in the worst case. We then propose a BE method for OPS, called Identifiable BE Selection (IBES), that has a straightforward method for selecting its own hyperparameters. We highlight that using IBES for OPS generally has more requirements than OPE methods, but if satisfied, can be more sample efficient. We conclude with an empirical study comparing OPE and IBES, and by showing the difficulty of OPS on an offline Atari benchmark dataset. △ Less

Submitted 4 December, 2023; originally announced December 2023.

GVFs in the Real World: Making Predictions Online for Water Treatment

Authors: Muhammad Kamran Janjua, Haseeb Shah, Martha White, Erfan Miahi, Marlos C. Machado, Adam White

Abstract: In this paper we investigate the use of reinforcement-learning based prediction approaches for a real drinking-water treatment plant. Develo** such a prediction system is a critical step on the path to optimizing and automating water treatment. Before that, there are many questions to answer about the predictability of the data, suitable neural network architectures, how to overcome partial obse… ▽ More In this paper we investigate the use of reinforcement-learning based prediction approaches for a real drinking-water treatment plant. Develo** such a prediction system is a critical step on the path to optimizing and automating water treatment. Before that, there are many questions to answer about the predictability of the data, suitable neural network architectures, how to overcome partial observability and more. We first describe this dataset, and highlight challenges with seasonality, nonstationarity, partial observability, and heterogeneity across sensors and operation modes of the plant. We then describe General Value Function (GVF) predictions -- discounted cumulative sums of observations -- and highlight why they might be preferable to classical n-step predictions common in time series prediction. We discuss how to use offline data to appropriately pre-train our temporal difference learning (TD) agents that learn these GVF predictions, including how to select hyperparameters for online fine-tuning in deployment. We find that the TD-prediction agent obtains an overall lower normalized mean-squared error than the n-step prediction agent. Finally, we show the importance of learning in deployment, by comparing a TD agent trained purely offline with no online updating to a TD agent that learns online. This final result is one of the first to motivate the importance of adapting predictions in real-time, for non-stationary high-volume systems in the real world. △ Less

Submitted 3 December, 2023; originally announced December 2023.

Comments: Published in Machine Learning (2023)

Journal ref: Machine Learning (2023): 1-31

Leveraging Artificial Intelligence Technology for Map** Research to Sustainable Development Goals: A Case Study

Authors: Hui Yin, Amir Aryani, Gavin Lambert, Marcus White, Luis Salvador-Carulla, Shazia Sadiq, Elvira Sojli, Jennifer Boddy, Greg Murray, Wing Wah Tham

Abstract: The number of publications related to the Sustainable Development Goals (SDGs) continues to grow. These publications cover a diverse spectrum of research, from humanities and social sciences to engineering and health. Given the imperative of funding bodies to monitor outcomes and impacts, linking publications to relevant SDGs is critical but remains time-consuming and difficult given the breadth a… ▽ More The number of publications related to the Sustainable Development Goals (SDGs) continues to grow. These publications cover a diverse spectrum of research, from humanities and social sciences to engineering and health. Given the imperative of funding bodies to monitor outcomes and impacts, linking publications to relevant SDGs is critical but remains time-consuming and difficult given the breadth and complexity of the SDGs. A publication may relate to several goals (interconnection feature of goals), and therefore require multidisciplinary knowledge to tag accurately. Machine learning approaches are promising and have proven particularly valuable for tasks such as manual data labeling and text classification. In this study, we employed over 82,000 publications from an Australian university as a case study. We utilized a similarity measure to map these publications onto Sustainable Development Goals (SDGs). Additionally, we leveraged the OpenAI GPT model to conduct the same task, facilitating a comparative analysis between the two approaches. Experimental results show that about 82.89% of the results obtained by the similarity measure overlap (at least one tag) with the outputs of the GPT model. The adopted model (similarity measure) can complement GPT model for SDG classification. Furthermore, deep learning methods, which include the similarity measure used here, are more accessible and trusted for dealing with sensitive data without the use of commercial AI services or the deployment of expensive computing resources to operate large language models. Our study demonstrates how a crafted combination of the two methods can achieve reliable results for map** research to the SDGs. △ Less

Submitted 9 November, 2023; originally announced November 2023.

ACM Class: I.2.7

Evaluating diversion and treatment policies for opioid use disorder

Authors: Veronica M. White, Laura A. Albert

Abstract: The United States opioid crisis contributed to 80,411 fatalities in 2021. It has strained hospitals, treatment facilities, and law enforcement agencies due to the enormous resources and procedures needed to respond to the crisis. As a result, many individuals who use opioids never receive or finish the treatment they need and instead have many interactions with hospitals or the criminal justice sy… ▽ More The United States opioid crisis contributed to 80,411 fatalities in 2021. It has strained hospitals, treatment facilities, and law enforcement agencies due to the enormous resources and procedures needed to respond to the crisis. As a result, many individuals who use opioids never receive or finish the treatment they need and instead have many interactions with hospitals or the criminal justice system. This paper introduces a discrete event simulation model that evaluates three opioid use disorder treatment policies: arrest diversion, re-entry case management, and overdose diversion. Publicly available data from 2011 to 2019 in Dane County, Wisconsin, was used to forecast opioid-related outcomes through 2032. Through analyzing a variety of policy-mix implementations, the study offers a versatile framework for evaluating policies at various implementation levels. The results demonstrate that treatment policies that create new pathways and programming by utilizing treatment services and successfully divert at least 20% of eligible individuals can lead to more opioid-resilient communities. The benefits increase when more policies are enacted and/or are offered to more individuals. We assume communities invest in increasing treatment capacity to meet increased treatment demand. In policy-mixes where societal savings from decreased opioid use, hospital encounters, and opioid-related arrests outweigh the costs of opioid use disorder treatment, the 2032 total savings range from $7.04 to $29.73 million. To reverse the opioid crisis within a community, treatment policies may need to be combined with other strategies, such as harm reduction, supply reduction, and use prevention. △ Less

Submitted 1 December, 2023; v1 submitted 8 November, 2023; originally announced November 2023.

Scalable Label-efficient Footpath Network Generation Using Remote Sensing Data and Self-supervised Learning

Authors: Xinye Wanyan, Sachith Seneviratne, Kerry Nice, Jason Thompson, Marcus White, Nano Langenheim, Mark Stevenson

Abstract: Footpath map**, modeling, and analysis can provide important geospatial insights to many fields of study, including transport, health, environment and urban planning. The availability of robust Geographic Information System (GIS) layers can benefit the management of infrastructure inventories, especially at local government level with urban planners responsible for the deployment and maintenance… ▽ More Footpath map**, modeling, and analysis can provide important geospatial insights to many fields of study, including transport, health, environment and urban planning. The availability of robust Geographic Information System (GIS) layers can benefit the management of infrastructure inventories, especially at local government level with urban planners responsible for the deployment and maintenance of such infrastructure. However, many cities still lack real-time information on the location, connectivity, and width of footpaths, and/or employ costly and manual survey means to gather this information. This work designs and implements an automatic pipeline for generating footpath networks based on remote sensing images using machine learning models. The annotation of segmentation tasks, especially labeling remote sensing images with specialized requirements, is very expensive, so we aim to introduce a pipeline requiring less labeled data. Considering supervised methods require large amounts of training data, we use a self-supervised method for feature representation learning to reduce annotation requirements. Then the pre-trained model is used as the encoder of the U-Net for footpath segmentation. Based on the generated masks, the footpath polygons are extracted and converted to footpath networks which can be loaded and visualized by geographic information systems conveniently. Validation results indicate considerable consistency when compared to manually collected GIS layers. The footpath network generation pipeline proposed in this work is low-cost and extensible, and it can be applied where remote sensing images are available. Github: https://github.com/WennyXY/FootpathSeg. △ Less

Submitted 17 September, 2023; originally announced September 2023.

Identifying depression-related topics in smartphone-collected free-response speech recordings using an automatic speech recognition system and a deep learning topic model

Authors: Yuezhou Zhang, Amos A Folarin, Judith Dineley, Pauline Conde, Valeria de Angel, Shaoxiong Sun, Yatharth Ranjan, Zulqarnain Rashid, Callum Stewart, Petroula Laiou, Heet Sankesara, Linglong Qian, Faith Matcham, Katie M White, Carolin Oetzmann, Femke Lamers, Sara Siddi, Sara Simblett, Björn W. Schuller, Srinivasan Vairavan, Til Wykes, Josep Maria Haro, Brenda WJH Penninx, Vaibhav A Narayan, Matthew Hotopf , et al. (3 additional authors not shown)

Abstract: Language use has been shown to correlate with depression, but large-scale validation is needed. Traditional methods like clinic studies are expensive. So, natural language processing has been employed on social media to predict depression, but limitations remain-lack of validated labels, biased user samples, and no context. Our study identified 29 topics in 3919 smartphone-collected speech recordi… ▽ More Language use has been shown to correlate with depression, but large-scale validation is needed. Traditional methods like clinic studies are expensive. So, natural language processing has been employed on social media to predict depression, but limitations remain-lack of validated labels, biased user samples, and no context. Our study identified 29 topics in 3919 smartphone-collected speech recordings from 265 participants using the Whisper tool and BERTopic model. Six topics with a median PHQ-8 greater than or equal to 10 were regarded as risk topics for depression: No Expectations, Sleep, Mental Therapy, Haircut, Studying, and Coursework. To elucidate the topic emergence and associations with depression, we compared behavioral (from wearables) and linguistic characteristics across identified topics. The correlation between topic shifts and changes in depression severity over time was also investigated, indicating the importance of longitudinally monitoring language use. We also tested the BERTopic model on a similar smaller dataset (356 speech recordings from 57 participants), obtaining some consistent results. In summary, our findings demonstrate specific speech topics may indicate depression severity. The presented data-driven workflow provides a practical approach to collecting and analyzing large-scale speech data from real-world settings for digital health research. △ Less

Submitted 5 September, 2023; v1 submitted 22 August, 2023; originally announced August 2023.

Measuring and Mitigating Interference in Reinforcement Learning

Authors: Vincent Liu, Han Wang, Ruo Yu Tao, Khurram Javed, Adam White, Martha White

Abstract: Catastrophic interference is common in many network-based learning systems, and many proposals exist for mitigating it. Before overcoming interference we must understand it better. In this work, we provide a definition and novel measure of interference for value-based reinforcement learning methods such as Fitted Q-Iteration and DQN. We systematically evaluate our measure of interference, showing… ▽ More Catastrophic interference is common in many network-based learning systems, and many proposals exist for mitigating it. Before overcoming interference we must understand it better. In this work, we provide a definition and novel measure of interference for value-based reinforcement learning methods such as Fitted Q-Iteration and DQN. We systematically evaluate our measure of interference, showing that it correlates with instability in control performance, across a variety of network architectures. Our new interference measure allows us to ask novel scientific questions about commonly used deep learning architectures and study learning algorithms which mitigate interference. Lastly, we outline a class of algorithms which we call online-aware that are designed to mitigate interference, and show they do reduce interference according to our measure and that they improve stability and performance in several classic control environments. △ Less

Submitted 10 July, 2023; originally announced July 2023.

Comments: Published at Conference on Lifelong Learning Agents (CoLLAs) 2023

Text-to-SQL Error Correction with Language Models of Code

Authors: Ziru Chen, Shijie Chen, Michael White, Raymond Mooney, Ali Payani, Jayanth Srinivasa, Yu Su, Huan Sun

Abstract: Despite recent progress in text-to-SQL parsing, current semantic parsers are still not accurate enough for practical use. In this paper, we investigate how to build automatic text-to-SQL error correction models. Noticing that token-level edits are out of context and sometimes ambiguous, we propose building clause-level edit models instead. Besides, while most language models of code are not specif… ▽ More Despite recent progress in text-to-SQL parsing, current semantic parsers are still not accurate enough for practical use. In this paper, we investigate how to build automatic text-to-SQL error correction models. Noticing that token-level edits are out of context and sometimes ambiguous, we propose building clause-level edit models instead. Besides, while most language models of code are not specifically pre-trained for SQL, they know common data structures and their operations in programming languages such as Python. Thus, we propose a novel representation for SQL queries and their edits that adheres more closely to the pre-training corpora of language models of code. Our error correction model improves the exact set match accuracy of different parsers by 2.4-6.5 and obtains up to 4.3 point absolute improvement over two strong baselines. Our code and data are available at https://github.com/OSU-NLP-Group/Auto-SQL-Correction. △ Less

Submitted 28 May, 2023; v1 submitted 22 May, 2023; originally announced May 2023.

Comments: ACL 2023 Short Paper

Coagent Networks: Generalized and Scaled

Authors: James E. Kostas, Scott M. Jordan, Yash Chandak, Georgios Theocharous, Dhawal Gupta, Martha White, Bruno Castro da Silva, Philip S. Thomas

Abstract: Coagent networks for reinforcement learning (RL) [Thomas and Barto, 2011] provide a powerful and flexible framework for deriving principled learning rules for arbitrary stochastic neural networks. The coagent framework offers an alternative to backpropagation-based deep learning (BDL) that overcomes some of backpropagation's main limitations. For example, coagent networks can compute different par… ▽ More Coagent networks for reinforcement learning (RL) [Thomas and Barto, 2011] provide a powerful and flexible framework for deriving principled learning rules for arbitrary stochastic neural networks. The coagent framework offers an alternative to backpropagation-based deep learning (BDL) that overcomes some of backpropagation's main limitations. For example, coagent networks can compute different parts of the network \emph{asynchronously} (at different rates or at different times), can incorporate non-differentiable components that cannot be used with backpropagation, and can explore at levels higher than their action spaces (that is, they can be designed as hierarchical networks for exploration and/or temporal abstraction). However, the coagent framework is not just an alternative to BDL; the two approaches can be blended: BDL can be combined with coagent learning rules to create architectures with the advantages of both approaches. This work generalizes the coagent theory and learning rules provided by previous works; this generalization provides more flexibility for network architecture design within the coagent framework. This work also studies one of the chief disadvantages of coagent networks: high variance updates for networks that have many coagents and do not use backpropagation. We show that a coagent algorithm with a policy network that does not use backpropagation can scale to a challenging RL domain with a high-dimensional state and action space (the MuJoCo Ant environment), learning reasonable (although not state-of-the-art) policies. These contributions motivate and provide a more general theoretical foundation for future work that studies coagent networks. △ Less

Submitted 16 May, 2023; originally announced May 2023.

Empirical Design in Reinforcement Learning

Authors: Andrew Patterson, Samuel Neumann, Martha White, Adam White

Abstract: Empirical design in reinforcement learning is no small task. Running good experiments requires attention to detail and at times significant computational resources. While compute resources available per dollar have continued to grow rapidly, so have the scale of typical experiments in reinforcement learning. It is now common to benchmark agents with millions of parameters against dozens of tasks,… ▽ More Empirical design in reinforcement learning is no small task. Running good experiments requires attention to detail and at times significant computational resources. While compute resources available per dollar have continued to grow rapidly, so have the scale of typical experiments in reinforcement learning. It is now common to benchmark agents with millions of parameters against dozens of tasks, each using the equivalent of 30 days of experience. The scale of these experiments often conflict with the need for proper statistical evidence, especially when comparing algorithms. Recent studies have highlighted how popular algorithms are sensitive to hyper-parameter settings and implementation details, and that common empirical practice leads to weak statistical evidence (Machado et al., 2018; Henderson et al., 2018). Here we take this one step further. This manuscript represents both a call to action, and a comprehensive resource for how to do good experiments in reinforcement learning. In particular, we cover: the statistical assumptions underlying common performance measures, how to properly characterize performance variation and stability, hypothesis testing, special considerations for comparing multiple agents, baseline and illustrative example construction, and how to deal with hyper-parameters and experimenter bias. Throughout we highlight common mistakes found in the literature and the statistical consequences of those in example experiments. The objective of this document is to provide answers on how we can use our unprecedented compute to do good science in reinforcement learning, as well as stay alert to potential pitfalls in our empirical design. △ Less

Submitted 3 April, 2023; originally announced April 2023.

Comments: In submission to JMLR

Exploring celebrity influence on public attitude towards the COVID-19 pandemic: social media shared sentiment analysis

Authors: Brianna M White, Chad A Melton, Parya Zareie, Robert L Davis, Robert A Bednarczyk, Arash Shaban-Nejad

Abstract: The COVID-19 pandemic has introduced new opportunities for health communication, including an increase in the public use of online outlets for health-related emotions. People have turned to social media networks to share sentiments related to the impacts of the COVID-19 pandemic. In this paper we examine the role of social messaging shared by Persons in the Public Eye (i.e. athletes, politicians,… ▽ More The COVID-19 pandemic has introduced new opportunities for health communication, including an increase in the public use of online outlets for health-related emotions. People have turned to social media networks to share sentiments related to the impacts of the COVID-19 pandemic. In this paper we examine the role of social messaging shared by Persons in the Public Eye (i.e. athletes, politicians, news personnel) in determining overall public discourse direction. We harvested approximately 13 million tweets ranging from 1 January 2020 to 1 March 2022. The sentiment was calculated for each tweet using a fine-tuned DistilRoBERTa model, which was used to compare COVID-19 vaccine-related Twitter posts (tweets) that co-occurred with mentions of People in the Public Eye. Our findings suggest the presence of consistent patterns of emotional content co-occurring with messaging shared by Persons in the Public Eye for the first two years of the COVID-19 pandemic influenced public opinion and largely stimulated online public discourse. We demonstrate that as the pandemic progressed, public sentiment shared on social networks was shaped by risk perceptions, political ideologies and health-protective behaviours shared by Persons in the Public Eye, often in a negative light. △ Less

Submitted 23 February, 2023; originally announced March 2023.

Comments: 7 Pages, 4 Figures

ACM Class: I.2.7

Journal ref: BMJ Health & Care Informatics 2023;30:e100665

The In-Sample Softmax for Offline Reinforcement Learning

Authors: Chenjun Xiao, Han Wang, Yangchen Pan, Adam White, Martha White

Abstract: Reinforcement learning (RL) agents can leverage batches of previously collected data to extract a reasonable control policy. An emerging issue in this offline RL setting, however, is that the bootstrap** update underlying many of our methods suffers from insufficient action-coverage: standard max operator may select a maximal action that has not been seen in the dataset. Bootstrap** from these… ▽ More Reinforcement learning (RL) agents can leverage batches of previously collected data to extract a reasonable control policy. An emerging issue in this offline RL setting, however, is that the bootstrap** update underlying many of our methods suffers from insufficient action-coverage: standard max operator may select a maximal action that has not been seen in the dataset. Bootstrap** from these inaccurate values can lead to overestimation and even divergence. There are a growing number of methods that attempt to approximate an \emph{in-sample} max, that only uses actions well-covered by the dataset. We highlight a simple fact: it is more straightforward to approximate an in-sample \emph{softmax} using only actions in the dataset. We show that policy iteration based on the in-sample softmax converges, and that for decreasing temperatures it approaches the in-sample max. We derive an In-Sample Actor-Critic (AC), using this in-sample softmax, and show that it is consistently better or comparable to existing offline RL methods, and is also well-suited to fine-tuning. △ Less

Submitted 19 April, 2023; v1 submitted 28 February, 2023; originally announced February 2023.

Asymptotically Unbiased Off-Policy Policy Evaluation when Reusing Old Data in Nonstationary Environments

Authors: Vincent Liu, Yash Chandak, Philip Thomas, Martha White

Abstract: In this work, we consider the off-policy policy evaluation problem for contextual bandits and finite horizon reinforcement learning in the nonstationary setting. Reusing old data is critical for policy evaluation, but existing estimators that reuse old data introduce large bias such that we can not obtain a valid confidence interval. Inspired from a related field called survey sampling, we introdu… ▽ More In this work, we consider the off-policy policy evaluation problem for contextual bandits and finite horizon reinforcement learning in the nonstationary setting. Reusing old data is critical for policy evaluation, but existing estimators that reuse old data introduce large bias such that we can not obtain a valid confidence interval. Inspired from a related field called survey sampling, we introduce a variant of the doubly robust (DR) estimator, called the regression-assisted DR estimator, that can incorporate the past data without introducing a large bias. The estimator unifies several existing off-policy policy evaluation methods and improves on them with the use of auxiliary information and a regression approach. We prove that the new estimator is asymptotically unbiased, and provide a consistent variance estimator to a construct a large sample confidence interval. Finally, we empirically show that the new estimator improves estimation for the current and future policy values, and provides a tight and valid interval estimation in several nonstationary recommendation environments. △ Less

Submitted 22 February, 2023; originally announced February 2023.

Comments: AISTATS 2023

Scalable Real-Time Recurrent Learning Using Columnar-Constructive Networks

Authors: Khurram Javed, Haseeb Shah, Rich Sutton, Martha White

Abstract: Constructing states from sequences of observations is an important component of reinforcement learning agents. One solution for state construction is to use recurrent neural networks. Back-propagation through time (BPTT), and real-time recurrent learning (RTRL) are two popular gradient-based methods for recurrent learning. BPTT requires complete trajectories of observations before it can compute t… ▽ More Constructing states from sequences of observations is an important component of reinforcement learning agents. One solution for state construction is to use recurrent neural networks. Back-propagation through time (BPTT), and real-time recurrent learning (RTRL) are two popular gradient-based methods for recurrent learning. BPTT requires complete trajectories of observations before it can compute the gradients and is unsuitable for online updates. RTRL can do online updates but scales poorly to large networks. In this paper, we propose two constraints that make RTRL scalable. We show that by either decomposing the network into independent modules or learning the network in stages, we can make RTRL scale linearly with the number of parameters. Unlike prior scalable gradient estimation algorithms, such as UORO and Truncated-BPTT, our algorithms do not add noise or bias to the gradient estimate. Instead, they trade off the functional capacity of the network for computationally efficient learning. We demonstrate the effectiveness of our approach over Truncated-BPTT on a prediction benchmark inspired by animal learning and by doing policy evaluation of pre-trained policies for Atari 2600 games. △ Less

Submitted 21 November, 2023; v1 submitted 20 January, 2023; originally announced February 2023.

Comments: Scalable recurrent learning, online learning, real-time recurrent learning, cascade correlation networks, agent-state construction, columnar networks, constructive networks

Generalized Munchausen Reinforcement Learning using Tsallis KL Divergence

Authors: Lingwei Zhu, Zheng Chen, Matthew Schlegel, Martha White

Abstract: Many policy optimization approaches in reinforcement learning incorporate a Kullback-Leilbler (KL) divergence to the previous policy, to prevent the policy from changing too quickly. This idea was initially proposed in a seminal paper on Conservative Policy Iteration, with approximations given by algorithms like TRPO and Munchausen Value Iteration (MVI). We continue this line of work by investigat… ▽ More Many policy optimization approaches in reinforcement learning incorporate a Kullback-Leilbler (KL) divergence to the previous policy, to prevent the policy from changing too quickly. This idea was initially proposed in a seminal paper on Conservative Policy Iteration, with approximations given by algorithms like TRPO and Munchausen Value Iteration (MVI). We continue this line of work by investigating a generalized KL divergence -- called the Tsallis KL divergence -- which use the $q$-logarithm in the definition. The approach is a strict generalization, as $q = 1$ corresponds to the standard KL divergence; $q > 1$ provides a range of new options. We characterize the types of policies learned under the Tsallis KL, and motivate when $q >1$ could be beneficial. To obtain a practical algorithm that incorporates Tsallis KL regularization, we extend MVI, which is one of the simplest approaches to incorporate KL regularization. We show that this generalized MVI($q$) obtains significant improvements over the standard MVI($q = 1$) across 35 Atari games. △ Less

Submitted 18 March, 2024; v1 submitted 26 January, 2023; originally announced January 2023.

Comments: Accepted by NeurIPS 2023

Trajectory-Aware Eligibility Traces for Off-Policy Reinforcement Learning

Authors: Brett Daley, Martha White, Christopher Amato, Marlos C. Machado

Abstract: Off-policy learning from multistep returns is crucial for sample-efficient reinforcement learning, but counteracting off-policy bias without exacerbating variance is challenging. Classically, off-policy bias is corrected in a per-decision manner: past temporal-difference errors are re-weighted by the instantaneous Importance Sampling (IS) ratio after each action via eligibility traces. Many off-po… ▽ More Off-policy learning from multistep returns is crucial for sample-efficient reinforcement learning, but counteracting off-policy bias without exacerbating variance is challenging. Classically, off-policy bias is corrected in a per-decision manner: past temporal-difference errors are re-weighted by the instantaneous Importance Sampling (IS) ratio after each action via eligibility traces. Many off-policy algorithms rely on this mechanism, along with differing protocols for cutting the IS ratios to combat the variance of the IS estimator. Unfortunately, once a trace has been fully cut, the effect cannot be reversed. This has led to the development of credit-assignment strategies that account for multiple past experiences at a time. These trajectory-aware methods have not been extensively analyzed, and their theoretical justification remains uncertain. In this paper, we propose a multistep operator that can express both per-decision and trajectory-aware methods. We prove convergence conditions for our operator in the tabular setting, establishing the first guarantees for several existing methods as well as many new ones. Finally, we introduce Recency-Bounded Importance Sampling (RBIS), which leverages trajectory awareness to perform robustly across $λ$-values in an off-policy control task. △ Less

Submitted 31 May, 2023; v1 submitted 26 January, 2023; originally announced January 2023.

Comments: ICML 2023. 8 pages, 2 figures. arXiv admin note: text overlap with arXiv:2112.12281

Fine-tuned Sentiment Analysis of COVID-19 Vaccine-Related Social Media Data: Comparative Study

Authors: Chad A Melton, Brianna M White, Robert L Davis, Robert A Bednarczyk, Arash Shaban-Nejad

Abstract: This study investigated and compared public sentiment related to COVID-19 vaccines expressed on two popular social media platforms, Reddit and Twitter, harvested from January 1, 2020, to March 1, 2022. To accomplish this task, we created a fine-tuned DistilRoBERTa model to predict sentiments of approximately 9.5 million Tweets and 70 thousand Reddit comments. To fine-tune our model, our team manua… ▽ More This study investigated and compared public sentiment related to COVID-19 vaccines expressed on two popular social media platforms, Reddit and Twitter, harvested from January 1, 2020, to March 1, 2022. To accomplish this task, we created a fine-tuned DistilRoBERTa model to predict sentiments of approximately 9.5 million Tweets and 70 thousand Reddit comments. To fine-tune our model, our team manually labeled the sentiment of 3600 Tweets and then augmented our dataset by the method of back-translation. Text sentiment for each social media platform was then classified with our fine-tuned model using Python and the Huggingface sentiment analysis pipeline. Our results determined that the average sentiment expressed on Twitter was more negative (52% positive) than positive and the sentiment expressed on Reddit was more positive than negative (53% positive). Though average sentiment was found to vary between these social media platforms, both displayed similar behavior related to sentiment shared at key vaccine-related developments during the pandemic. Considering this similar trend in shared sentiment demonstrated across social media platforms, Twitter and Reddit continue to be valuable data sources that public health officials can utilize to strengthen vaccine confidence and combat misinformation. As the spread of misinformation poses a range of psychological and psychosocial risks (anxiety, fear, etc.), there is an urgency in understanding the public perspective and attitude toward shared falsities. Comprehensive educational delivery systems tailored to the population's expressed sentiments that facilitate digital literacy, health information-seeking behavior, and precision health promotion could aid in clarifying such misinformation. △ Less

Submitted 17 October, 2022; originally announced November 2022.

Comments: 11 Pages, 5 Figures, and 1 Table

MSC Class: 92-11 ACM Class: I.2.7

Journal ref: Journal of Medical Internet Research (JMIR) 2022;24(10):e40408

Probing of Quantitative Values in Abstractive Summarization Models

Authors: Nathan M. White

Abstract: Abstractive text summarization has recently become a popular approach, but data hallucination remains a serious problem, including with quantitative data. We propose a set of probing tests to evaluate the efficacy of abstract summarization models' modeling of quantitative values found in the input text. Our results show that in most cases, the encoders of recent SOTA-performing models struggle to… ▽ More Abstractive text summarization has recently become a popular approach, but data hallucination remains a serious problem, including with quantitative data. We propose a set of probing tests to evaluate the efficacy of abstract summarization models' modeling of quantitative values found in the input text. Our results show that in most cases, the encoders of recent SOTA-performing models struggle to provide embeddings that adequately represent quantitative values in the input compared to baselines, and in particular, they outperform random representations in some, but surprisingly not all, cases. Under our assumptions, this suggests that the encoder's performance contributes to the quantity hallucination problem. One model type in particular, DistilBART-CDM, was observed to underperform randomly initialized representations for several experiments, and performance versus BERT suggests that standard pretraining and fine-tuning approaches for the summarization task may play a role in underperformance for some encoders. △ Less

Submitted 2 October, 2022; originally announced October 2022.

Comments: 9 pages

GEMv2: Multilingual NLG Benchmarking in a Single Line of Code

Authors: Sebastian Gehrmann, Abhik Bhattacharjee, Abinaya Mahendiran, Alex Wang, Alexandros Papangelis, Aman Madaan, Angelina McMillan-Major, Anna Shvets, Ashish Upadhyay, Bingsheng Yao, Bryan Wilie, Chandra Bhagavatula, Chaobin You, Craig Thomson, Cristina Garbacea, Dakuo Wang, Daniel Deutsch, Deyi Xiong, Di **, Dimitra Gkatzia, Dragomir Radev, Elizabeth Clark, Esin Durmus, Faisal Ladhak, Filip Ginter , et al. (52 additional authors not shown)

Abstract: Evaluation in machine learning is usually informed by past choices, for example which datasets or metrics to use. This standardization enables the comparison on equal footing using leaderboards, but the evaluation choices become sub-optimal as better alternatives arise. This problem is especially pertinent in natural language generation which requires ever-improving suites of datasets, metrics, an… ▽ More Evaluation in machine learning is usually informed by past choices, for example which datasets or metrics to use. This standardization enables the comparison on equal footing using leaderboards, but the evaluation choices become sub-optimal as better alternatives arise. This problem is especially pertinent in natural language generation which requires ever-improving suites of datasets, metrics, and human evaluation to make definitive claims. To make following best model evaluation practices easier, we introduce GEMv2. The new version of the Generation, Evaluation, and Metrics Benchmark introduces a modular infrastructure for dataset, model, and metric developers to benefit from each others work. GEMv2 supports 40 documented datasets in 51 languages. Models for all datasets can be evaluated online and our interactive data card creation and rendering tools make it easier to add new datasets to the living benchmark. △ Less

Submitted 24 June, 2022; v1 submitted 22 June, 2022; originally announced June 2022.

Goal-Space Planning with Subgoal Models

Authors: Chunlok Lo, Kevin Roice, Parham Mohammad Panahi, Scott Jordan, Adam White, Gabor Mihucz, Farzane Aminmansour, Martha White

Abstract: This paper investigates a new approach to model-based reinforcement learning using background planning: mixing (approximate) dynamic programming updates and model-free updates, similar to the Dyna architecture. Background planning with learned models is often worse than model-free alternatives, such as Double DQN, even though the former uses significantly more memory and computation. The fundament… ▽ More This paper investigates a new approach to model-based reinforcement learning using background planning: mixing (approximate) dynamic programming updates and model-free updates, similar to the Dyna architecture. Background planning with learned models is often worse than model-free alternatives, such as Double DQN, even though the former uses significantly more memory and computation. The fundamental problem is that learned models can be inaccurate and often generate invalid states, especially when iterated many steps. In this paper, we avoid this limitation by constraining background planning to a set of (abstract) subgoals and learning only local, subgoal-conditioned models. This goal-space planning (GSP) approach is more computationally efficient, naturally incorporates temporal abstraction for faster long-horizon planning and avoids learning the transition dynamics entirely. We show that our GSP algorithm can propagate value from an abstract space in a manner that helps a variety of base learners learn significantly faster in different domains. △ Less

Submitted 27 February, 2024; v1 submitted 6 June, 2022; originally announced June 2022.

No More Pesky Hyperparameters: Offline Hyperparameter Tuning for RL

Authors: Han Wang, Archit Sakhadeo, Adam White, James Bell, Vincent Liu, Xutong Zhao, Puer Liu, Tadashi Kozuno, Alona Fyshe, Martha White

Abstract: The performance of reinforcement learning (RL) agents is sensitive to the choice of hyperparameters. In real-world settings like robotics or industrial control systems, however, testing different hyperparameter configurations directly on the environment can be financially prohibitive, dangerous, or time consuming. We propose a new approach to tune hyperparameters from offline logs of data, to full… ▽ More The performance of reinforcement learning (RL) agents is sensitive to the choice of hyperparameters. In real-world settings like robotics or industrial control systems, however, testing different hyperparameter configurations directly on the environment can be financially prohibitive, dangerous, or time consuming. We propose a new approach to tune hyperparameters from offline logs of data, to fully specify the hyperparameters for an RL agent that learns online in the real world. The approach is conceptually simple: we first learn a model of the environment from the offline data, which we call a calibration model, and then simulate learning in the calibration model to identify promising hyperparameters. We identify several criteria to make this strategy effective, and develop an approach that satisfies these criteria. We empirically investigate the method in a variety of settings to identify when it is effective and when it fails. △ Less

Submitted 18 May, 2022; originally announced May 2022.

Robust Losses for Learning Value Functions

Authors: Andrew Patterson, Victor Liao, Martha White

Abstract: Most value function learning algorithms in reinforcement learning are based on the mean squared (projected) Bellman error. However, squared errors are known to be sensitive to outliers, both skewing the solution of the objective and resulting in high-magnitude and high-variance gradients. To control these high-magnitude updates, typical strategies in RL involve clip** gradients, clip** rewards… ▽ More Most value function learning algorithms in reinforcement learning are based on the mean squared (projected) Bellman error. However, squared errors are known to be sensitive to outliers, both skewing the solution of the objective and resulting in high-magnitude and high-variance gradients. To control these high-magnitude updates, typical strategies in RL involve clip** gradients, clip** rewards, rescaling rewards, or clip** errors. While these strategies appear to be related to robust losses -- like the Huber loss -- they are built on semi-gradient update rules which do not minimize a known loss. In this work, we build on recent insights reformulating squared Bellman errors as a saddlepoint optimization problem and propose a saddlepoint reformulation for a Huber Bellman error and Absolute Bellman error. We start from a formalization of robust losses, then derive sound gradient-based approaches to minimize these losses in both the online off-policy prediction and control settings. We characterize the solutions of the robust losses, providing insight into the problem settings where the robust losses define notably better solutions than the mean squared Bellman error. Finally, we show that the resulting gradient-based algorithms are more stable, for both prediction and control, with less sensitivity to meta-parameters. △ Less

Submitted 17 April, 2023; v1 submitted 17 May, 2022; originally announced May 2022.

Comments: IEEE Transactions on Pattern Analysis and Machine Intelligence (2022)

Searching for Efficient Neural Architectures for On-Device ML on Edge TPUs

Authors: Berkin Akin, Suyog Gupta, Yun Long, Anton Spiridonov, Zhuo Wang, Marie White, Hao Xu, ** Zhou, Yanqi Zhou

Abstract: On-device ML accelerators are becoming a standard in modern mobile system-on-chips (SoC). Neural architecture search (NAS) comes to the rescue for efficiently utilizing the high compute throughput offered by these accelerators. However, existing NAS frameworks have several practical limitations in scaling to multiple tasks and different target platforms. In this work, we provide a two-pronged appr… ▽ More On-device ML accelerators are becoming a standard in modern mobile system-on-chips (SoC). Neural architecture search (NAS) comes to the rescue for efficiently utilizing the high compute throughput offered by these accelerators. However, existing NAS frameworks have several practical limitations in scaling to multiple tasks and different target platforms. In this work, we provide a two-pronged approach to this challenge: (i) a NAS-enabling infrastructure that decouples model cost evaluation, search space design, and the NAS algorithm to rapidly target various on-device ML tasks, and (ii) search spaces crafted from group convolution based inverted bottleneck (IBN) variants that provide flexible quality/performance trade-offs on ML accelerators, complementing the existing full and depthwise convolution based IBNs. Using this approach we target a state-of-the-art mobile platform, Google Tensor SoC, and demonstrate neural architectures that improve the quality-performance pareto frontier for various computer vision (classification, detection, segmentation) as well as natural language processing tasks. △ Less

Submitted 8 April, 2022; originally announced April 2022.

FastMapSVM: Classifying Complex Objects Using the FastMap Algorithm and Support-Vector Machines

Authors: Malcolm C. A. White, Kushal Sharma, Ang Li, T. K. Satish Kumar, Nori Nakata

Abstract: Neural Networks and related Deep Learning methods are currently at the leading edge of technologies used for classifying objects. However, they generally demand large amounts of time and data for model training; and their learned models can sometimes be difficult to interpret. In this paper, we advance FastMapSVM -- an interpretable Machine Learning framework for classifying complex objects -- as… ▽ More Neural Networks and related Deep Learning methods are currently at the leading edge of technologies used for classifying objects. However, they generally demand large amounts of time and data for model training; and their learned models can sometimes be difficult to interpret. In this paper, we advance FastMapSVM -- an interpretable Machine Learning framework for classifying complex objects -- as an advantageous alternative to Neural Networks for general classification tasks. FastMapSVM extends the applicability of Support-Vector Machines (SVMs) to domains with complex objects by combining the complementary strengths of FastMap and SVMs. FastMap is an efficient linear-time algorithm that maps complex objects to points in a Euclidean space while preserving pairwise domain-specific distances between them. We demonstrate the efficiency and effectiveness of FastMapSVM in the context of classifying seismograms. We show that its performance, in terms of precision, recall, and accuracy, is comparable to that of other state-of-the-art methods. However, compared to other methods, FastMapSVM uses significantly smaller amounts of time and data for model training. It also provides a perspicuous visualization of the objects and the classification boundaries between them. We expect FastMapSVM to be viable for classification tasks in many other real-world domains. △ Less

Submitted 15 June, 2022; v1 submitted 7 April, 2022; originally announced April 2022.

Comments: 27 pages, 12 figures

Investigating the Properties of Neural Network Representations in Reinforcement Learning

Authors: Han Wang, Erfan Miahi, Martha White, Marlos C. Machado, Zaheer Abbas, Raksha Kumaraswamy, Vincent Liu, Adam White

Abstract: In this paper we investigate the properties of representations learned by deep reinforcement learning systems. Much of the early work on representations for reinforcement learning focused on designing fixed-basis architectures to achieve properties thought to be desirable, such as orthogonality and sparsity. In contrast, the idea behind deep reinforcement learning methods is that the agent designe… ▽ More In this paper we investigate the properties of representations learned by deep reinforcement learning systems. Much of the early work on representations for reinforcement learning focused on designing fixed-basis architectures to achieve properties thought to be desirable, such as orthogonality and sparsity. In contrast, the idea behind deep reinforcement learning methods is that the agent designer should not encode representational properties, but rather that the data stream should determine the properties of the representation -- good representations emerge under appropriate training schemes. In this paper we bring these two perspectives together, empirically investigating the properties of representations that support transfer in reinforcement learning. We introduce and measure six representational properties over more than 25 thousand agent-task settings. We consider Deep Q-learning agents with different auxiliary losses in a pixel-based navigation environment, with source and transfer tasks corresponding to different goal locations. We develop a method to better understand why some representations work better for transfer, through a systematic approach varying task similarity and measuring and correlating representation properties with transfer performance. We demonstrate the generality of the methodology by investigating representations learned by a Rainbow agent that successfully transfer across games modes in Atari 2600. △ Less

Submitted 5 May, 2023; v1 submitted 29 March, 2022; originally announced March 2022.

Digital Fingerprinting of Microstructures

Authors: Michael D. White, Alexander Tarakanov, Christopher P. Race, Philip J. Withers, Kody J. H. Law

Abstract: Finding efficient means of fingerprinting microstructural information is a critical step towards harnessing data-centric machine learning approaches. A statistical framework is systematically developed for compressed characterisation of a population of images, which includes some classical computer vision methods as special cases. The focus is on materials microstructure. The ultimate purpose is t… ▽ More Finding efficient means of fingerprinting microstructural information is a critical step towards harnessing data-centric machine learning approaches. A statistical framework is systematically developed for compressed characterisation of a population of images, which includes some classical computer vision methods as special cases. The focus is on materials microstructure. The ultimate purpose is to rapidly fingerprint sample images in the context of various high-throughput design/make/test scenarios. This includes, but is not limited to, quantification of the disparity between microstructures for quality control, classifying microstructures, predicting materials properties from image data and identifying potential processing routes to engineer new materials with specific properties. Here, we consider microstructure classification and utilise the resulting features over a range of related machine learning tasks, namely supervised, semi-supervised, and unsupervised learning. The approach is applied to two distinct datasets to illustrate various aspects and some recommendations are made based on the findings. In particular, methods that leverage transfer learning with convolutional neural networks (CNNs), pretrained on the ImageNet dataset, are generally shown to outperform other methods. Additionally, dimensionality reduction of these CNN-based fingerprints is shown to have negligible impact on classification accuracy for the supervised learning approaches considered. In situations where there is a large dataset with only a handful of images labelled, graph-based label propagation to unlabelled data is shown to be favourable over discarding unlabelled data and performing supervised learning. In particular, label propagation by Poisson learning is shown to be highly effective at low label rates. △ Less

Submitted 22 January, 2024; v1 submitted 25 March, 2022; originally announced March 2022.

Resonance in Weight Space: Covariate Shift Can Drive Divergence of SGD with Momentum

Authors: Kirby Banman, Liam Peet-Pare, Nidhi Hegde, Alona Fyshe, Martha White

Abstract: Most convergence guarantees for stochastic gradient descent with momentum (SGDm) rely on iid sampling. Yet, SGDm is often used outside this regime, in settings with temporally correlated input samples such as continual learning and reinforcement learning. Existing work has shown that SGDm with a decaying step-size can converge under Markovian temporal correlation. In this work, we show that SGDm u… ▽ More Most convergence guarantees for stochastic gradient descent with momentum (SGDm) rely on iid sampling. Yet, SGDm is often used outside this regime, in settings with temporally correlated input samples such as continual learning and reinforcement learning. Existing work has shown that SGDm with a decaying step-size can converge under Markovian temporal correlation. In this work, we show that SGDm under covariate shift with a fixed step-size can be unstable and diverge. In particular, we show SGDm under covariate shift is a parametric oscillator, and so can suffer from a phenomenon known as resonance. We approximate the learning system as a time varying system of ordinary differential equations, and leverage existing theory to characterize the system's divergence/convergence as resonant/nonresonant modes. The theoretical result is limited to the linear setting with periodic covariate shift, so we empirically supplement this result to show that resonance phenomena persist even under non-periodic covariate shift, nonlinear dynamics with neural networks, and optimizers other than SGDm. △ Less

Submitted 22 March, 2022; originally announced March 2022.

Comments: In International Conference on Learning Representations. 2021

Continual Auxiliary Task Learning

Authors: Matthew McLeod, Chunlok Lo, Matthew Schlegel, Andrew Jacobsen, Raksha Kumaraswamy, Martha White, Adam White

Abstract: Learning auxiliary tasks, such as multiple predictions about the world, can provide many benefits to reinforcement learning systems. A variety of off-policy learning algorithms have been developed to learn such predictions, but as yet there is little work on how to adapt the behavior to gather useful data for those off-policy predictions. In this work, we investigate a reinforcement learning syste… ▽ More Learning auxiliary tasks, such as multiple predictions about the world, can provide many benefits to reinforcement learning systems. A variety of off-policy learning algorithms have been developed to learn such predictions, but as yet there is little work on how to adapt the behavior to gather useful data for those off-policy predictions. In this work, we investigate a reinforcement learning system designed to learn a collection of auxiliary tasks, with a behavior policy learning to take actions to improve those auxiliary predictions. We highlight the inherent non-stationarity in this continual auxiliary task learning problem, for both prediction learners and the behavior learner. We develop an algorithm based on successor features that facilitates tracking under non-stationary rewards, and prove the separation into learning successor features and rewards provides convergence rate improvements. We conduct an in-depth study into the resulting multi-prediction learning system. △ Less

Submitted 22 February, 2022; originally announced February 2022.

Comments: Neural Information Processing Systems 2021

A Temporal-Difference Approach to Policy Gradient Estimation

Authors: Samuele Tosatto, Andrew Patterson, Martha White, A. Rupam Mahmood

Abstract: The policy gradient theorem (Sutton et al., 2000) prescribes the usage of a cumulative discounted state distribution under the target policy to approximate the gradient. Most algorithms based on this theorem, in practice, break this assumption, introducing a distribution shift that can cause the convergence to poor solutions. In this paper, we propose a new approach of reconstructing the policy gr… ▽ More The policy gradient theorem (Sutton et al., 2000) prescribes the usage of a cumulative discounted state distribution under the target policy to approximate the gradient. Most algorithms based on this theorem, in practice, break this assumption, introducing a distribution shift that can cause the convergence to poor solutions. In this paper, we propose a new approach of reconstructing the policy gradient from the start state without requiring a particular sampling strategy. The policy gradient calculation in this form can be simplified in terms of a gradient critic, which can be recursively estimated due to a new Bellman equation of gradients. By using temporal-difference updates of the gradient critic from an off-policy data stream, we develop the first estimator that sidesteps the distribution shift issue in a model-free way. We prove that, under certain realizability conditions, our estimator is unbiased regardless of the sampling strategy. We empirically show that our technique achieves a superior bias-variance trade-off and performance in presence of off-policy samples. △ Less

Submitted 7 July, 2022; v1 submitted 4 February, 2022; originally announced February 2022.

An Alternate Policy Gradient Estimator for Softmax Policies

Authors: Shivam Garg, Samuele Tosatto, Yangchen Pan, Martha White, A. Rupam Mahmood

Abstract: Policy gradient (PG) estimators are ineffective in dealing with softmax policies that are sub-optimally saturated, which refers to the situation when the policy concentrates its probability mass on sub-optimal actions. Sub-optimal policy saturation may arise from bad policy initialization or sudden changes in the environment that occur after the policy has already converged. Current softmax PG est… ▽ More Policy gradient (PG) estimators are ineffective in dealing with softmax policies that are sub-optimally saturated, which refers to the situation when the policy concentrates its probability mass on sub-optimal actions. Sub-optimal policy saturation may arise from bad policy initialization or sudden changes in the environment that occur after the policy has already converged. Current softmax PG estimators require a large number of updates to overcome policy saturation, which causes low sample efficiency and poor adaptability to new situations. To mitigate this problem, we propose a novel PG estimator for softmax policies that utilizes the bias in the critic estimate and the noise present in the reward signal to escape the saturated regions of the policy parameter space. Our theoretical analysis and experiments, conducted on bandits and various reinforcement learning environments, show that this new estimator is significantly more robust to policy saturation. △ Less

Submitted 24 February, 2022; v1 submitted 21 December, 2021; originally announced December 2021.

Comments: Accepted to AISTATS 2022. 60 pages, 50 figures. This updated version has an additional experiment and minor corrections

Representation Alignment in Neural Networks

Authors: Ehsan Imani, Wei Hu, Martha White

Abstract: It is now a standard for neural network representations to be trained on large, publicly available datasets, and used for new problems. The reasons for why neural network representations have been so successful for transfer, however, are still not fully understood. In this paper we show that, after training, neural network representations align their top singular vectors to the targets. We investi… ▽ More It is now a standard for neural network representations to be trained on large, publicly available datasets, and used for new problems. The reasons for why neural network representations have been so successful for transfer, however, are still not fully understood. In this paper we show that, after training, neural network representations align their top singular vectors to the targets. We investigate this representation alignment phenomenon in a variety of neural network architectures and find that (a) alignment emerges across a variety of different architectures and optimizers, with more alignment arising from depth (b) alignment increases for layers closer to the output and (c) existing high-performance deep CNNs exhibit high levels of alignment. We then highlight why alignment between the top singular vectors and the targets can speed up learning and show in a classic synthetic transfer problem that representation alignment correlates with positive and negative transfer to similar and dissimilar tasks. △ Less

Submitted 17 September, 2022; v1 submitted 14 December, 2021; originally announced December 2021.

Comments: 26 pages, 21 figures

Off-Policy Actor-Critic with Emphatic Weightings

Authors: Eric Graves, Ehsan Imani, Raksha Kumaraswamy, Martha White

Abstract: A variety of theoretically-sound policy gradient algorithms exist for the on-policy setting due to the policy gradient theorem, which provides a simplified form for the gradient. The off-policy setting, however, has been less clear due to the existence of multiple objectives and the lack of an explicit off-policy policy gradient theorem. In this work, we unify these objectives into one off-policy… ▽ More A variety of theoretically-sound policy gradient algorithms exist for the on-policy setting due to the policy gradient theorem, which provides a simplified form for the gradient. The off-policy setting, however, has been less clear due to the existence of multiple objectives and the lack of an explicit off-policy policy gradient theorem. In this work, we unify these objectives into one off-policy objective, and provide a policy gradient theorem for this unified objective. The derivation involves emphatic weightings and interest functions. We show multiple strategies to approximate the gradients, in an algorithm called Actor Critic with Emphatic weightings (ACE). We prove in a counterexample that previous (semi-gradient) off-policy actor-critic methods--particularly Off-Policy Actor-Critic (OffPAC) and Deterministic Policy Gradient (DPG)--converge to the wrong solution whereas ACE finds the optimal solution. We also highlight why these semi-gradient approaches can still perform well in practice, suggesting strategies for variance reduction in ACE. We empirically study several variants of ACE on two classic control environments and an image-based environment designed to illustrate the tradeoffs made by each gradient approximation. We find that by approximating the emphatic weightings directly, ACE performs as well as or better than OffPAC in all settings tested. △ Less

Submitted 13 April, 2023; v1 submitted 15 November, 2021; originally announced November 2021.

Comments: 63 pages

Journal ref: Journal of Machine Learning Research 24 (2023) 1-63

Exploiting Action Impact Regularity and Exogenous State Variables for Offline Reinforcement Learning

Authors: Vincent Liu, James R. Wright, Martha White

Abstract: Offline reinforcement learning -- learning a policy from a batch of data -- is known to be hard for general MDPs. These results motivate the need to look at specific classes of MDPs where offline reinforcement learning might be feasible. In this work, we explore a restricted class of MDPs to obtain guarantees for offline reinforcement learning. The key property, which we call Action Impact Regular… ▽ More Offline reinforcement learning -- learning a policy from a batch of data -- is known to be hard for general MDPs. These results motivate the need to look at specific classes of MDPs where offline reinforcement learning might be feasible. In this work, we explore a restricted class of MDPs to obtain guarantees for offline reinforcement learning. The key property, which we call Action Impact Regularity (AIR), is that actions primarily impact a part of the state (an endogenous component) and have limited impact on the remaining part of the state (an exogenous component). AIR is a strong assumption, but it nonetheless holds in a number of real-world domains including financial markets. We discuss algorithms that exploit the AIR property, and provide a theoretical analysis for an algorithm based on Fitted-Q Iteration. Finally, we demonstrate that the algorithm outperforms existing offline reinforcement learning algorithms across different data collection policies in simulated and real world environments where the regularity holds. △ Less

Submitted 3 May, 2023; v1 submitted 15 November, 2021; originally announced November 2021.

Journal ref: Journal of Artificial Intelligence Research, 77 (2023) 71-101

Towards Transparent Interactive Semantic Parsing via Step-by-Step Correction

Authors: Lingbo Mo, Ashley Lewis, Huan Sun, Michael White

Abstract: Existing studies on semantic parsing focus primarily on map** a natural-language utterance to a corresponding logical form in one turn. However, because natural language can contain a great deal of ambiguity and variability, this is a difficult challenge. In this work, we investigate an interactive semantic parsing framework that explains the predicted logical form step by step in natural langua… ▽ More Existing studies on semantic parsing focus primarily on map** a natural-language utterance to a corresponding logical form in one turn. However, because natural language can contain a great deal of ambiguity and variability, this is a difficult challenge. In this work, we investigate an interactive semantic parsing framework that explains the predicted logical form step by step in natural language and enables the user to make corrections through natural-language feedback for individual steps. We focus on question answering over knowledge bases (KBQA) as an instantiation of our framework, aiming to increase the transparency of the parsing process and help the user appropriately trust the final answer. To do so, we construct INSPIRED, a crowdsourced dialogue dataset derived from the ComplexWebQuestions dataset. Our experiments show that the interactive framework with human feedback has the potential to greatly improve overall parse accuracy. Furthermore, we develop a pipeline for dialogue simulation to evaluate our framework w.r.t. a variety of state-of-the-art KBQA models without involving further crowdsourcing effort. The results demonstrate that our interactive semantic parsing framework promises to be effective across such models. △ Less

Submitted 27 March, 2022; v1 submitted 15 October, 2021; originally announced October 2021.

Comments: Accepted by Findings of ACL 2022

When are Deep Networks really better than Decision Forests at small sample sizes, and how?

Authors: Haoyin Xu, Kaleab A. Kinfu, Will LeVine, Sambit Panda, Jayanta Dey, Michael Ainsworth, Yu-Chung Peng, Madi Kusmanov, Florian Engert, Christopher M. White, Joshua T. Vogelstein, Carey E. Priebe

Abstract: Deep networks and decision forests (such as random forests and gradient boosted trees) are the leading machine learning methods for structured and tabular data, respectively. Many papers have empirically compared large numbers of classifiers on one or two different domains (e.g., on 100 different tabular data settings). However, a careful conceptual and empirical comparison of these two strategies… ▽ More Deep networks and decision forests (such as random forests and gradient boosted trees) are the leading machine learning methods for structured and tabular data, respectively. Many papers have empirically compared large numbers of classifiers on one or two different domains (e.g., on 100 different tabular data settings). However, a careful conceptual and empirical comparison of these two strategies using the most contemporary best practices has yet to be performed. Conceptually, we illustrate that both can be profitably viewed as "partition and vote" schemes. Specifically, the representation space that they both learn is a partitioning of feature space into a union of convex polytopes. For inference, each decides on the basis of votes from the activated nodes. This formulation allows for a unified basic understanding of the relationship between these methods. Empirically, we compare these two strategies on hundreds of tabular data settings, as well as several vision and auditory settings. Our focus is on datasets with at most 10,000 samples, which represent a large fraction of scientific and biomedical datasets. In general, we found forests to excel at tabular and structured data (vision and audition) with small sample sizes, whereas deep nets performed better on structured data with larger sample sizes. This suggests that further gains in both scenarios may be realized via further combining aspects of forests and networks. We will continue revising this technical report in the coming months with updated results. △ Less

Submitted 2 November, 2021; v1 submitted 31 August, 2021; originally announced August 2021.

Bespoke Fractal Sampling Patterns for Discrete Fourier Space via the Kaleidoscope Transform

Authors: Jacob M. White, Stuart Crozier, Shekhar S. Chandra

Abstract: Sampling strategies are important for sparse imaging methodologies, especially those employing the discrete Fourier transform (DFT). Chaotic sensing is one such methodology that employs deterministic, fractal sampling in conjunction with finite, iterative reconstruction schemes to form an image from limited samples. Using a sampling pattern constructed entirely from periodic lines in DFT space, ch… ▽ More Sampling strategies are important for sparse imaging methodologies, especially those employing the discrete Fourier transform (DFT). Chaotic sensing is one such methodology that employs deterministic, fractal sampling in conjunction with finite, iterative reconstruction schemes to form an image from limited samples. Using a sampling pattern constructed entirely from periodic lines in DFT space, chaotic sensing was found to outperform traditional compressed sensing for magnetic resonance imaging; however, only one such sampling pattern was presented and the reason for its fractal nature was not proven. Through the introduction of a novel image transform known as the kaleidoscope transform, which formalises and extends upon the concept of downsampling and concatenating an image with itself, this paper: (1) demonstrates a fundamental relationship between multiplication in modular arithmetic and downsampling; (2) provides a rigorous mathematical explanation for the fractal nature of the sampling pattern in the DFT; and (3) leverages this understanding to develop a collection of novel fractal sampling patterns for the 2D DFT with customisable properties. The ability to design tailor-made fractal sampling patterns expands the utility of the DFT in chaotic imaging and may form the basis for a bespoke chaotic sensing methodology, in which the fractal sampling matches the imaging task for improved reconstruction. △ Less

Submitted 2 August, 2021; originally announced August 2021.

Comments: 6 pages, 7 figures

Greedification Operators for Policy Optimization: Investigating Forward and Reverse KL Divergences

Authors: Alan Chan, Hugo Silva, Sungsu Lim, Tadashi Kozuno, A. Rupam Mahmood, Martha White

Abstract: Approximate Policy Iteration (API) algorithms alternate between (approximate) policy evaluation and (approximate) greedification. Many different approaches have been explored for approximate policy evaluation, but less is understood about approximate greedification and what choices guarantee policy improvement. In this work, we investigate approximate greedification when reducing the KL divergence… ▽ More Approximate Policy Iteration (API) algorithms alternate between (approximate) policy evaluation and (approximate) greedification. Many different approaches have been explored for approximate policy evaluation, but less is understood about approximate greedification and what choices guarantee policy improvement. In this work, we investigate approximate greedification when reducing the KL divergence between the parameterized policy and the Boltzmann distribution over action values. In particular, we investigate the difference between the forward and reverse KL divergences, with varying degrees of entropy regularization. We show that the reverse KL has stronger policy improvement guarantees, but that reducing the forward KL can result in a worse policy. We also demonstrate, however, that a large enough reduction of the forward KL can induce improvement under additional assumptions. Empirically, we show on simple continuous-action environments that the forward KL can induce more exploration, but at the cost of a more suboptimal policy. No significant differences were observed in the discrete-action setting or on a suite of benchmark problems. Throughout, we highlight that many policy gradient methods can be seen as an instance of API, with either the forward or reverse KL for the policy update, and discuss next steps for understanding and improving our policy optimization algorithms. △ Less

Submitted 18 April, 2022; v1 submitted 17 July, 2021; originally announced July 2021.

Comments: Updated the paper with more theory in Section 5 and moved some experiments to the Appendix

Leveraging semantically similar queries for ranking via combining representations

Authors: Hayden S. Helm, Marah Abdin, Benjamin D. Pedigo, Shweti Mahajan, Vince Lyzinski, Youngser Park, Amitabh Basu, Piali~Choudhury, Christopher M. White, Weiwei Yang, Carey E. Priebe

Abstract: In modern ranking problems, different and disparate representations of the items to be ranked are often available. It is sensible, then, to try to combine these representations to improve ranking. Indeed, learning to rank via combining representations is both principled and practical for learning a ranking function for a particular query. In extremely data-scarce settings, however, the amount of l… ▽ More In modern ranking problems, different and disparate representations of the items to be ranked are often available. It is sensible, then, to try to combine these representations to improve ranking. Indeed, learning to rank via combining representations is both principled and practical for learning a ranking function for a particular query. In extremely data-scarce settings, however, the amount of labeled data available for a particular query can lead to a highly variable and ineffective ranking function. One way to mitigate the effect of the small amount of data is to leverage information from semantically similar queries. Indeed, as we demonstrate in simulation settings and real data examples, when semantically similar queries are available it is possible to gainfully use them when ranking with respect to a particular query. We describe and explore this phenomenon in the context of the bias-variance trade off and apply it to the data-scarce settings of a Bing navigational graph and the Drosophila larva connectome. △ Less

Submitted 23 June, 2021; originally announced June 2021.

Predictive Representation Learning for Language Modeling

Authors: Qingfeng Lan, Luke Kumar, Martha White, Alona Fyshe

Abstract: To effectively perform the task of next-word prediction, long short-term memory networks (LSTMs) must keep track of many types of information. Some information is directly related to the next word's identity, but some is more secondary (e.g. discourse-level features or features of downstream words). Correlates of secondary information appear in LSTM representations even though they are not part of… ▽ More To effectively perform the task of next-word prediction, long short-term memory networks (LSTMs) must keep track of many types of information. Some information is directly related to the next word's identity, but some is more secondary (e.g. discourse-level features or features of downstream words). Correlates of secondary information appear in LSTM representations even though they are not part of an \emph{explicitly} supervised prediction task. In contrast, in reinforcement learning (RL), techniques that explicitly supervise representations to predict secondary information have been shown to be beneficial. Inspired by that success, we propose Predictive Representation Learning (PRL), which explicitly constrains LSTMs to encode specific predictions, like those that might need to be learned implicitly. We show that PRL 1) significantly improves two strong language modeling methods, 2) converges more quickly, and 3) performs better when data is limited. Our work shows that explicitly encoding a simple predictive task facilitates the search for a more effective language model. △ Less

Submitted 29 May, 2021; originally announced May 2021.