ISIT 2022

204 papers
The 2022 IEEE International Symposium on Information Theory (ISIT) featured 238 papers addressing core information-theoretic problems and their applications in machine learning. A prominent research theme involved generalization in deep learning, with studies focusing on computable generalization bounds for self-certified learning, neural collapse under cross-entropy loss, and the behavior of overparameterized models. Optimization techniques also received attention, specifically Bregman proximal methods and strategies for reducing gradient variance in stochastic processes. Another area of inquiry was partial information decomposition for analyzing complex, multimodal, and multivariate systems. The year's contributions also covered advancements in coding theory and signal processing, including list decoding of binary alternant codes, universal 1-bit compressive sensing, and universal error correction using foundation models. Research further explored practical communication constraints, such as finite block-length regimes and joint communication and privacy requirements. Prolific authors for this edition included Yihan Zhang, Bo Bai, Antonia Wachter-Zeh, and Deniz Gündüz. The proceedings reflected a continued integration of information theory with federated learning and the mitigation of spurious correlations in large-scale models.

Top Research Directions

Research directions most represented by ISIT 2022 papers.

Geometry Aware Optimization via Relative Smoothness 2
Straggler-Resilient Coded Computation 2
Moment Constraints for Information Measures on Infinite Alphabets 1
Probabilistic Constellation Shaping 1
Learning from Dependent Data with Unknown Mixing Rates 1
Tightening the Bounds Between Total Variation and KL Divergence 1
Asymptotically Optimal Quickest Change Detection 1
Monolithic Neural Architectures for End to End Wireless Receivers 1
Sequential Rademacher Complexity for Online Learning 1
The Duality Between Information and Estimation 1
Exploiting Statistical Correlations in Group Testing 1
Bandwidth Adaptive Repair for Distributed Storage 1
Non-Uniform Noise Modeling in Ranking and Sorting 1
Quantifying Generalization via Information Leakage 1
Regret Bounds for Gaussian Process Bandits 1

Top Authors

AuthorPapers
Yihan Zhang7
Antonia Wachter-Zeh6
Deniz Gunduz6
Alex Dytso5
Bo Bai5
Martina Cardone5
Cheuk Ting Li4
Christina Fragouli4
Jonathan Scarlett4
Rawad Bitar4
Chao Tian3
Christos Thrampoulidis3
Farhad Shirani3
Gong Zhang3
Hamed Hassani3
Oliver Kosut3
Songze Li3
Vincent Y. F. Tan3
Alfred O. Hero2
Anand D. Sarwate2

Impactful Papers

Papers from ISIT 2022 that shaped research clusters.

Foundational

Influential

Papers

Title
Precise Minimax Regret for Logistic Regression
Privacy Leakage in Discrete-Time Updating Systems
Privacy Limits in Power-Law Bipartite Networks under Active Fingerprinting Attacks
Proactive Resilience in 1-2-1 Networks
Protecting Semantic Information Using An Efficient Secret Key
Proving Information Inequalities and Identities with Symbolic Computation
Query Age of Information: Optimizing AoI at the Right Time
Regularization-wise double descent: Why it occurs and how to eliminate it
Remote Contextual Bandits
Robust Estimation for Non-parametric Families via Generative Adversarial Networks
Robust Locally Positioning Sequences and Codes: Capacity, Constructions and Applications
Robust Multi-Hypothesis Testing with Moment-Constrained Uncertainty Sets
Run Length Limited de Bruijn Sequences for Quantum Communications
Scheduling Group Tests over Time
Secure Coding via Gaussian Random Fields
Secure Distributed/Federated Learning: Prediction-Privacy Trade-Off for Multi-Agent System
Semi-Decentralized Federated Learning with Collaborative Relaying
Sensitivity of Under-Determined Linear System
Sequential vs. Fixed Design Regrets in Online Learning
Simple Binary Hypothesis Testing under Communication Constraints
Sketching sparse low-rank matrices with near-optimal sample- and time-complexity
Social Learning under Randomized Collaborations
Sparse superposition codes under VAMP decoding with generic rotational invariant coding matrices
Statistical Delay and Error-Rate Bounded QoS Control for URLLC in the Non-Asymptotic Regime
Statistical Minimax Lower Bounds for Transfer Learning in Linear Binary Classification
Statistical QoS-Driven Beamforming and Trajectory Optimizations in UAV/IRS-Based 6G Wireless Networks in the Non-Asymptotic Regime
Stochastic Chaining and Strengthened Information-Theoretic Generalization Bounds
Stochastic Coded Federated Learning with Convergence and Privacy Guarantees
SwiftAgg: Communication-Efficient and Dropout-Resistant Secure Aggregation for Federated Learning with Worst-Case Security Guarantees
The Capacity of Causal Adversarial Channels
The Compound Information Bottleneck Program
The Directional Bias Helps Stochastic Gradient Descent to Generalize in Kernel Regression Models
The Dissemination of Time-Varying Information over Networked Agents with Gossiping
The Gapped k-Deck Problem
The Generalized Degrees-of-Freedom of the Asymmetric Interference Channel with Delayed CSIT
The Optimal Sample Complexity of Matrix Completion with Hierarchical Similarity Graphs
The Quantum MAC with Cribbing Encoders
The Random Number Partitioning Problem: Overlap Gap Property and Algorithmic Barriers
The Undecidability of Network Coding with some Fixed-Size Messages and Edges
Tighter Expected Generalization Error Bounds via Convexity of Information Measures
TinyTurbo: Efficient Turbo Decoders on Edge
Towards Efficient Repair and Coding of Binary MDS Array Codes with Small Sub-packetization
Understanding Deep Neural Networks Using Sliced Mutual Information
Understanding Generalization via Leave-One-Out Conditional Mutual Information
Uniqueness of local maximizers for some non-convex log-determinant optimization problems using information theory
Universal 1-Bit Compressive Sensing for Bounded Dynamic Range Signals
Universal Compression of Large Alphabets with Constrained Compressors
Upper Bounds on the Feedback Error Exponent of Channels With States and With Memory
Upper Confidence Interval Strategies for Multi-Armed Bandits with Entropy Rewards
Using One Redundant Bit to Construct Two-Dimensional Almost-Balanced Codes

« prev Page 4/5 next »