-
Automatic Prediction of Amyotrophic Lateral Sclerosis Progression using Longitudinal Speech Transformer
Authors:
Liming Wang,
Yuan Gong,
Nauman Dawalatabad,
Marco Vilela,
Katerina Placek,
Brian Tracey,
Yishu Gong,
Alan Premasiri,
Fernando Vieira,
James Glass
Abstract:
Automatic prediction of amyotrophic lateral sclerosis (ALS) disease progression provides a more efficient and objective alternative than manual approaches. We propose ALS longitudinal speech transformer (ALST), a neural network-based automatic predictor of ALS disease progression from longitudinal speech recordings of ALS patients. By taking advantage of high-quality pretrained speech features and…
▽ More
Automatic prediction of amyotrophic lateral sclerosis (ALS) disease progression provides a more efficient and objective alternative than manual approaches. We propose ALS longitudinal speech transformer (ALST), a neural network-based automatic predictor of ALS disease progression from longitudinal speech recordings of ALS patients. By taking advantage of high-quality pretrained speech features and longitudinal information in the recordings, our best model achieves 91.0\% AUC, improving upon the previous best model by 5.6\% relative on the ALS TDI dataset. Careful analysis reveals that ALST is capable of fine-grained and interpretable predictions of ALS progression, especially for distinguishing between rarer and more severe cases. Code is publicly available.
△ Less
Submitted 26 June, 2024;
originally announced June 2024.
-
Evaluating Cognitive Maps and Planning in Large Language Models with CogEval
Authors:
Ida Momennejad,
Hosein Hasanbeig,
Felipe Vieira,
Hiteshi Sharma,
Robert Osazuwa Ness,
Nebojsa Jojic,
Hamid Palangi,
Jonathan Larson
Abstract:
Recently an influx of studies claim emergent cognitive abilities in large language models (LLMs). Yet, most rely on anecdotes, overlook contamination of training sets, or lack systematic Evaluation involving multiple tasks, control conditions, multiple iterations, and statistical robustness tests. Here we make two major contributions. First, we propose CogEval, a cognitive science-inspired protoco…
▽ More
Recently an influx of studies claim emergent cognitive abilities in large language models (LLMs). Yet, most rely on anecdotes, overlook contamination of training sets, or lack systematic Evaluation involving multiple tasks, control conditions, multiple iterations, and statistical robustness tests. Here we make two major contributions. First, we propose CogEval, a cognitive science-inspired protocol for the systematic evaluation of cognitive capacities in Large Language Models. The CogEval protocol can be followed for the evaluation of various abilities. Second, here we follow CogEval to systematically evaluate cognitive maps and planning ability across eight LLMs (OpenAI GPT-4, GPT-3.5-turbo-175B, davinci-003-175B, Google Bard, Cohere-xlarge-52.4B, Anthropic Claude-1-52B, LLaMA-13B, and Alpaca-7B). We base our task prompts on human experiments, which offer both established construct validity for evaluating planning, and are absent from LLM training sets. We find that, while LLMs show apparent competence in a few planning tasks with simpler structures, systematic evaluation reveals striking failure modes in planning tasks, including hallucinations of invalid trajectories and getting trapped in loops. These findings do not support the idea of emergent out-of-the-box planning ability in LLMs. This could be because LLMs do not understand the latent relational structures underlying planning problems, known as cognitive maps, and fail at unrolling goal-directed trajectories based on the underlying structure. Implications for application and future directions are discussed.
△ Less
Submitted 24 September, 2023;
originally announced September 2023.
-
A Bayesian actor-oriented multilevel relational event model with hypothesis testing procedures
Authors:
Fabio Vieira,
Roger Leenders,
Daniel McFarland,
Joris Mulder
Abstract:
Relational event network data are becoming increasingly available. Consequently, statistical models for such data have also surfaced. These models mainly focus on the analysis of single networks, while in many applications, multiple independent event sequences are observed, which are likely to display similar social interaction dynamics. Furthermore, statistical methods for testing hypotheses abou…
▽ More
Relational event network data are becoming increasingly available. Consequently, statistical models for such data have also surfaced. These models mainly focus on the analysis of single networks, while in many applications, multiple independent event sequences are observed, which are likely to display similar social interaction dynamics. Furthermore, statistical methods for testing hypotheses about social interaction behavior are underdeveloped. Therefore, the contribution of the current paper is twofold. First, we present a multilevel extension of the dynamic actor-oriented model, which allows researchers to model sender and receiver processes separately. The multilevel formulation enables principled probabilistic borrowing of information across networks to accurately estimate drivers of social dynamics. Second, a flexible methodology is proposed to test hypotheses about common and heterogeneous social interaction drivers across relational event sequences. Social interaction data between children and teachers in classrooms are used to showcase the methodology.
△ Less
Submitted 7 June, 2023; v1 submitted 22 April, 2022;
originally announced April 2022.
-
BlockColdChain: Vaccine Cold Chain Blockchain
Authors:
Ronan D. Mendonça,
Otávio S. Gomes,
Luiz F. M. Vieira,
Marcos A. M. Vieira,
Alex B. Vieira,
José A. M. Nacif
Abstract:
In this paper, we propose a blockchain-based cold chain technology for vaccine cooling track. The COVID-19 pandemic has caused the death of millions of people. An important step towards ending the pandemic is vaccination. Vaccines must be kept under control temperature during the whole process, from fabrication to the hands of the health professionals who will immunize the population. However, the…
▽ More
In this paper, we propose a blockchain-based cold chain technology for vaccine cooling track. The COVID-19 pandemic has caused the death of millions of people. An important step towards ending the pandemic is vaccination. Vaccines must be kept under control temperature during the whole process, from fabrication to the hands of the health professionals who will immunize the population. However, there are numerous reports of vaccine loss due to temperature variations, and, currently, people getting vaccinated have no control if their vaccine was kept safe. Blockchain is a technology solution that can provide public and verifiable records. We review the World Health Organization (WHO) cool chain and Blockchain technology. Moreover, we describe current IoT temperature monitoring devices and propose Blockcoldchain to track vaccine cold chain using blockchain, thus proving an unalterable vaccine temperature history. Our experimental results using smart contracts demonstrate the system's feasibility.
△ Less
Submitted 28 April, 2021;
originally announced April 2021.
-
Entropy as a measure of attractiveness and socioeconomic complexity in Rio de Janeiro metropolitan area
Authors:
Maxime Lenormand,
Horacio Samaniego,
Julio C. Chaves,
Vinicius F. Vieira,
Moacyr A. H. B. da Silva,
Alexandre G. Evsukoff
Abstract:
Defining and measuring spatial inequalities across the urban environment remains a complex and elusive task that has been facilitated by the increasing availability of large geolocated databases. In this study, we rely on a mobile phone dataset and an entropy-based metric to measure the attractiveness of a location in the Rio de Janeiro Metropolitan Area (Brazil) as the diversity of visitors' loca…
▽ More
Defining and measuring spatial inequalities across the urban environment remains a complex and elusive task that has been facilitated by the increasing availability of large geolocated databases. In this study, we rely on a mobile phone dataset and an entropy-based metric to measure the attractiveness of a location in the Rio de Janeiro Metropolitan Area (Brazil) as the diversity of visitors' location of residence. The results show that the attractiveness of a given location measured by entropy is an important descriptor of the socioeconomic status of the location, and can thus be used as a proxy for complex socioeconomic indicators.
△ Less
Submitted 23 March, 2020;
originally announced March 2020.
-
Personalizing ASR for Dysarthric and Accented Speech with Limited Data
Authors:
Joel Shor,
Dotan Emanuel,
Oran Lang,
Omry Tuval,
Michael Brenner,
Julie Cattiau,
Fernando Vieira,
Maeve McNally,
Taylor Charbonneau,
Melissa Nollstadt,
Avinatan Hassidim,
Yossi Matias
Abstract:
Automatic speech recognition (ASR) systems have dramatically improved over the last few years. ASR systems are most often trained from 'typical' speech, which means that underrepresented groups don't experience the same level of improvement. In this paper, we present and evaluate finetuning techniques to improve ASR for users with non-standard speech. We focus on two types of non-standard speech:…
▽ More
Automatic speech recognition (ASR) systems have dramatically improved over the last few years. ASR systems are most often trained from 'typical' speech, which means that underrepresented groups don't experience the same level of improvement. In this paper, we present and evaluate finetuning techniques to improve ASR for users with non-standard speech. We focus on two types of non-standard speech: speech from people with amyotrophic lateral sclerosis (ALS) and accented speech. We train personalized models that achieve 62% and 35% relative WER improvement on these two groups, bringing the absolute WER for ALS speakers, on a test set of message bank phrases, down to 10% for mild dysarthria and 20% for more serious dysarthria. We show that 71% of the improvement comes from only 5 minutes of training data. Finetuning a particular subset of layers (with many fewer parameters) often gives better results than finetuning the entire model. This is the first step towards building state of the art ASR models for dysarthric speech.
△ Less
Submitted 31 July, 2019;
originally announced July 2019.
-
Devito: Towards a generic Finite Difference DSL using Symbolic Python
Authors:
Michael Lange,
Navjot Kukreja,
Mathias Louboutin,
Fabio Luporini,
Felippe Vieira,
Vincenzo Pandolfo,
Paulius Velesko,
Paulius Kazakas,
Gerard Gorman
Abstract:
Domain specific languages (DSL) have been used in a variety of fields to express complex scientific problems in a concise manner and provide automated performance optimization for a range of computational architectures. As such DSLs provide a powerful mechanism to speed up scientific Python computation that goes beyond traditional vectorization and pre-compilation approaches, while allowing domain…
▽ More
Domain specific languages (DSL) have been used in a variety of fields to express complex scientific problems in a concise manner and provide automated performance optimization for a range of computational architectures. As such DSLs provide a powerful mechanism to speed up scientific Python computation that goes beyond traditional vectorization and pre-compilation approaches, while allowing domain scientists to build applications within the comforts of the Python software ecosystem. In this paper we present Devito, a new finite difference DSL that provides optimized stencil computation from high-level problem specifications based on symbolic Python expressions. We demonstrate Devito's symbolic API and performance advantages over traditional Python acceleration methods before highlighting its use in the scientific context of seismic inversion problems.
△ Less
Submitted 12 September, 2016;
originally announced September 2016.
-
Devito: automated fast finite difference computation
Authors:
Navjot Kukreja,
Mathias Louboutin,
Felippe Vieira,
Fabio Luporini,
Michael Lange,
Gerard Gorman
Abstract:
Domain specific languages have successfully been used in a variety of fields to cleanly express scientific problems as well as to simplify implementation and performance opti- mization on different computer architectures. Although a large number of stencil languages are available, finite differ- ence domain specific languages have proved challenging to design because most practical use cases requi…
▽ More
Domain specific languages have successfully been used in a variety of fields to cleanly express scientific problems as well as to simplify implementation and performance opti- mization on different computer architectures. Although a large number of stencil languages are available, finite differ- ence domain specific languages have proved challenging to design because most practical use cases require additional features that fall outside the finite difference abstraction. Inspired by the complexity of real-world seismic imaging problems, we introduce Devito, a domain specific language in which high level equations are expressed using symbolic expressions from the SymPy package. Complex equations are automatically manipulated, optimized, and translated into highly optimized C code that aims to perform compa- rably or better than hand-tuned code. All this is transpar- ent to users, who only see concise symbolic mathematical expressions.
△ Less
Submitted 10 October, 2016; v1 submitted 30 August, 2016;
originally announced August 2016.
-
Scheduling wireless links by graph multicoloring in the physical interference model
Authors:
Fabio R. J. Vieira,
José F. de Rezende,
Valmir C. Barbosa
Abstract:
Scheduling wireless links for simultaneous activation in such a way that all transmissions are successfully decoded at the receivers and moreover network capacity is maximized is a computationally hard problem. Usually it is tackled by heuristics whose output is a sequence of time slots in which every link appears in exactly one time slot. Such approaches can be interpreted as the coloring of a gr…
▽ More
Scheduling wireless links for simultaneous activation in such a way that all transmissions are successfully decoded at the receivers and moreover network capacity is maximized is a computationally hard problem. Usually it is tackled by heuristics whose output is a sequence of time slots in which every link appears in exactly one time slot. Such approaches can be interpreted as the coloring of a graph's vertices so that every vertex gets exactly one color. Here we introduce a new approach that can be viewed as assigning multiple colors to each vertex, so that, in the resulting schedule, every link may appear more than once (though the same number of times for all links). We report on extensive computational experiments, under the physical interference model, revealing substantial gains for a variety of randomly generated networks.
△ Less
Submitted 7 April, 2015;
originally announced April 2015.
-
Local heuristic for the refinement of multi-path routing in wireless mesh networks
Authors:
Fabio R. J. Vieira,
José F. de Rezende,
Valmir C. Barbosa,
Serge Fdida
Abstract:
We consider wireless mesh networks and the problem of routing end-to-end traffic over multiple paths for the same origin-destination pair with minimal interference. We introduce a heuristic for path determination with two distinguishing characteristics. First, it works by refining an extant set of paths, determined previously by a single- or multi-path routing algorithm. Second, it is totally loca…
▽ More
We consider wireless mesh networks and the problem of routing end-to-end traffic over multiple paths for the same origin-destination pair with minimal interference. We introduce a heuristic for path determination with two distinguishing characteristics. First, it works by refining an extant set of paths, determined previously by a single- or multi-path routing algorithm. Second, it is totally local, in the sense that it can be run by each of the origins on information that is available no farther than the node's immediate neighborhood. We have conducted extensive computational experiments with the new heuristic, using AODV and OLSR, as well as their multi-path variants, as underlying routing methods. For two different CSMA settings (as implemented by 802.11) and one TDMA setting running a path-oriented link scheduling algorithm, we have demonstrated that the new heuristic is capable of improving the average throughput network-wide. When working from the paths generated by the multi-path routing algorithms, the heuristic is also capable to provide a more evenly distributed traffic pattern.
△ Less
Submitted 8 March, 2012;
originally announced March 2012.
-
Scheduling links for heavy traffic on interfering routes in wireless mesh networks
Authors:
Fabio R. J. Vieira,
José F. de Rezende,
Valmir C. Barbosa,
Serge Fdida
Abstract:
We consider wireless mesh networks and the problem of scheduling the links of a given set of routes under the assumption of a heavy-traffic pattern. We assume some TDMA protocol provides a background of synchronized time slots and seek to schedule the routes' links to maximize the number of packets that get delivered to their destinations per time slot. Our approach is to construct an undirected g…
▽ More
We consider wireless mesh networks and the problem of scheduling the links of a given set of routes under the assumption of a heavy-traffic pattern. We assume some TDMA protocol provides a background of synchronized time slots and seek to schedule the routes' links to maximize the number of packets that get delivered to their destinations per time slot. Our approach is to construct an undirected graph G and to heuristically obtain node multicolorings for G that can be turned into efficient link schedules. In G each node represents a link to be scheduled and the edges are set up to represent every possible interference for any given set of interference assumptions. We present two multicoloring-based heuristics and study their performance through extensive simulations. One of the two heuristics is based on relaxing the notion of a node multicoloring by dynamically exploiting the availability of communication opportunities that would otherwise be wasted. We have found that, as a consequence, its performance is significantly superior to the other's.
△ Less
Submitted 8 June, 2011;
originally announced June 2011.
-
Optimization of supply diversity for the self-assembly of simple objects in two and three dimensions
Authors:
Fabio R. J. Vieira,
Valmir C. Barbosa
Abstract:
The field of algorithmic self-assembly is concerned with the design and analysis of self-assembly systems from a computational perspective, that is, from the perspective of mathematical problems whose study may give insight into the natural processes through which elementary objects self-assemble into more complex ones. One of the main problems of algorithmic self-assembly is the minimum tile se…
▽ More
The field of algorithmic self-assembly is concerned with the design and analysis of self-assembly systems from a computational perspective, that is, from the perspective of mathematical problems whose study may give insight into the natural processes through which elementary objects self-assemble into more complex ones. One of the main problems of algorithmic self-assembly is the minimum tile set problem (MTSP), which asks for a collection of types of elementary objects (called tiles) to be found for the self-assembly of an object having a pre-established shape. Such a collection is to be as concise as possible, thus minimizing supply diversity, while satisfying a set of stringent constraints having to do with the termination and other properties of the self-assembly process from its tile types. We present a study of what we think is the first practical approach to MTSP. Our study starts with the introduction of an evolutionary heuristic to tackle MTSP and includes results from extensive experimentation with the heuristic on the self-assembly of simple objects in two and three dimensions. The heuristic we introduce combines classic elements from the field of evolutionary computation with a problem-specific variant of Pareto dominance into a multi-objective approach to MTSP.
△ Less
Submitted 4 October, 2007; v1 submitted 3 October, 2007;
originally announced October 2007.
