-
A multiscale cavity method for sublinear-rank symmetric matrix factorization
Authors:
Jean Barbier,
Justin Ko,
Anas A. Rahman
Abstract:
We consider a statistical model for symmetric matrix factorization with additive Gaussian noise in the high-dimensional regime where the rank $M$ of the signal matrix to infer scales with its size $N$ as $M = o(N^{1/10})$. Allowing for a $N$-dependent rank offers new challenges and requires new methods. Working in the Bayesian-optimal setting, we show that whenever the signal has i.i.d. entries th…
▽ More
We consider a statistical model for symmetric matrix factorization with additive Gaussian noise in the high-dimensional regime where the rank $M$ of the signal matrix to infer scales with its size $N$ as $M = o(N^{1/10})$. Allowing for a $N$-dependent rank offers new challenges and requires new methods. Working in the Bayesian-optimal setting, we show that whenever the signal has i.i.d. entries the limiting mutual information between signal and data is given by a variational formula involving a rank-one replica symmetric potential. In other words, from the information-theoretic perspective, the case of a (slowly) growing rank is the same as when $M = 1$ (namely, the standard spiked Wigner model). The proof is primarily based on a novel multiscale cavity method allowing for growing rank along with some information-theoretic identities on worst noise for the Gaussian vector channel. We believe that the cavity method developed here will play a role in the analysis of a broader class of inference and spin models where the degrees of freedom are large arrays instead of vectors.
△ Less
Submitted 11 March, 2024;
originally announced March 2024.
-
Transformers in Reinforcement Learning: A Survey
Authors:
Pranav Agarwal,
Aamer Abdul Rahman,
Pierre-Luc St-Charles,
Simon J. D. Prince,
Samira Ebrahimi Kahou
Abstract:
Transformers have significantly impacted domains like natural language processing, computer vision, and robotics, where they improve performance compared to other neural networks. This survey explores how transformers are used in reinforcement learning (RL), where they are seen as a promising solution for addressing challenges such as unstable training, credit assignment, lack of interpretability,…
▽ More
Transformers have significantly impacted domains like natural language processing, computer vision, and robotics, where they improve performance compared to other neural networks. This survey explores how transformers are used in reinforcement learning (RL), where they are seen as a promising solution for addressing challenges such as unstable training, credit assignment, lack of interpretability, and partial observability. We begin by providing a brief domain overview of RL, followed by a discussion on the challenges of classical RL algorithms. Next, we delve into the properties of the transformer and its variants and discuss the characteristics that make them well-suited to address the challenges inherent in RL. We examine the application of transformers to various aspects of RL, including representation learning, transition and reward function modeling, and policy optimization. We also discuss recent research that aims to enhance the interpretability and efficiency of transformers in RL, using visualization techniques and efficient training strategies. Often, the transformer architecture must be tailored to the specific needs of a given application. We present a broad overview of how transformers have been adapted for several applications, including robotics, medicine, language modeling, cloud computing, and combinatorial optimization. We conclude by discussing the limitations of using transformers in RL and assess their potential for catalyzing future breakthroughs in this field.
△ Less
Submitted 12 July, 2023;
originally announced July 2023.
-
Pitfalls of Conditional Batch Normalization for Contextual Multi-Modal Learning
Authors:
Ivaxi Sheth,
Aamer Abdul Rahman,
Mohammad Havaei,
Samira Ebrahimi Kahou
Abstract:
Humans have perfected the art of learning from multiple modalities through sensory organs. Despite their impressive predictive performance on a single modality, neural networks cannot reach human level accuracy with respect to multiple modalities. This is a particularly challenging task due to variations in the structure of respective modalities. Conditional Batch Normalization (CBN) is a popular…
▽ More
Humans have perfected the art of learning from multiple modalities through sensory organs. Despite their impressive predictive performance on a single modality, neural networks cannot reach human level accuracy with respect to multiple modalities. This is a particularly challenging task due to variations in the structure of respective modalities. Conditional Batch Normalization (CBN) is a popular method that was proposed to learn contextual features to aid deep learning tasks. This technique uses auxiliary data to improve representational power by learning affine transformations for convolutional neural networks. Despite the boost in performance observed by using CBN layers, our work reveals that the visual features learned by introducing auxiliary data via CBN deteriorates. We perform comprehensive experiments to evaluate the brittleness of CBN networks to various datasets, suggesting that learning from visual features alone could often be superior for generalization. We evaluate CBN models on natural images for bird classification and histology images for cancer type classification. We observe that the CBN network learns close to no visual features on the bird classification dataset and partial visual features on the histology dataset. Our extensive experiments reveal that CBN may encourage shortcut learning between the auxiliary data and labels.
△ Less
Submitted 28 November, 2022;
originally announced November 2022.
-
Formal Verification of Safety Properties Using Interpolation and k-induction
Authors:
Tephilla Prince,
Atif Abdur Rahman,
Sheerazuddin Syed
Abstract:
This technical report presents implementation of two symbolic model checking algorithms that use SAT/SMT Solvers, namely interpolation based model checking and k-induction based model checking. We also do a comparative analysis of these two model checking algorithms.
This technical report presents implementation of two symbolic model checking algorithms that use SAT/SMT Solvers, namely interpolation based model checking and k-induction based model checking. We also do a comparative analysis of these two model checking algorithms.
△ Less
Submitted 4 July, 2022;
originally announced July 2022.
-
Virtualization Implementation Model for Cost Effective & Efficient Data Centers
Authors:
Mueen Uddin,
Azizah Abdul Rahman
Abstract:
Data centers form a key part of the infrastructure upon which a variety of information technology services are built. They provide the capabilities of centralized repository for storage, management, networking and dissemination of data. With the rapid increase in the capacity and size of data centers, there is a continuous increase in the demand for energy consumption. These data centers not only…
▽ More
Data centers form a key part of the infrastructure upon which a variety of information technology services are built. They provide the capabilities of centralized repository for storage, management, networking and dissemination of data. With the rapid increase in the capacity and size of data centers, there is a continuous increase in the demand for energy consumption. These data centers not only consume a tremendous amount of energy but are riddled with IT inefficiencies. Data center are plagued with thousands of servers as major components. These servers consume huge energy without performing useful work. In an average server environment, 30% of the servers are "dead" only consuming energy, without being properly utilized. This paper proposes a five step model using an emerging technology called virtualization to achieve energy efficient data centers. The proposed model helps Data Center managers to properly implement virtualization technology in their data centers to make them green and energy efficient so as to ensure that IT infrastructure contributes as little as possible to the emission of greenhouse gases, and helps to regain power and cooling capacity, recapture resilience and dramatically reducing energy costs and total cost of ownership.
△ Less
Submitted 7 April, 2012;
originally announced June 2012.
-
Server Consolidation: An Approach to make Data Centers Energy Efficient and Green
Authors:
Mueen Uddin,
Azizah Abdul Rahman
Abstract:
Data centers are the building blocks of IT business organizations providing the capabilities of centralized repository for storage, management, networking and dissemination of data. With the rapid increase in the capacity and size of data centers, there is a continuous increase in the demand for energy consumption. These data centers not only consume a tremendous amount of energy but are riddled w…
▽ More
Data centers are the building blocks of IT business organizations providing the capabilities of centralized repository for storage, management, networking and dissemination of data. With the rapid increase in the capacity and size of data centers, there is a continuous increase in the demand for energy consumption. These data centers not only consume a tremendous amount of energy but are riddled with IT inefficiencies. All data center are plagued with thousands of servers as major components. These servers consume huge energy without performing useful work. In an average server environment, 30% of the servers are "dead" only consuming energy, without being properly utilized. Their utilization ratio is only 5 to 10 percent. This paper focuses on the use of an emerging technology called virtualization to achieve energy efficient data centers by providing a solution called server consolidation. It increases the utilization ratio up to 50% saving huge amount of energy. Server consolidation helps in implementing green data centers to ensure that IT infrastructure contributes as little as possible to the emission of green house gases, and helps to regain power and cooling capacity, recapture resilience and dramatically reducing energy costs and total cost of ownership.
△ Less
Submitted 25 October, 2010;
originally announced October 2010.
-
Dynamic Multi Layer Signature based Intrusion Detection system Using Mobile Agents
Authors:
Mueen Uddin,
Azizah Abdul Rahman
Abstract:
Intrusion detection systems have become a key component in ensuring the safety of systems and networks. As networks grow in size and speed continues to increase, it is crucial that efficient scalable techniques should be developed for IDS systems. Signature based detection is the most extensively used threat detection technique for Intrusion Detection Systems (IDS). One of the foremost challenges…
▽ More
Intrusion detection systems have become a key component in ensuring the safety of systems and networks. As networks grow in size and speed continues to increase, it is crucial that efficient scalable techniques should be developed for IDS systems. Signature based detection is the most extensively used threat detection technique for Intrusion Detection Systems (IDS). One of the foremost challenges for signaturebased IDS systems is how to keep up with large volume of incoming traffic when each packet needs to be compared with every signature in the database. When an IDS cannot keep up with the traffic flood, all it can do is to drop packets, therefore, may miss potential attacks. This paper proposes a new model called Dynamic Multi-Layer Signature based IDS using Mobile Agents, which can detect imminent threats with very high success rate by dynamically and automatically creating and using small and efficient multiple databases, and at the same time, provide mechanism to update these small signature databases at regular intervals using Mobile Agents
△ Less
Submitted 25 October, 2010;
originally announced October 2010.
