Asilomar 2026 || Pacific Grove, California || October 25

TP8a: Deep Learning, Bandits, and Applications

Tue, 27 Oct, 13:30 - 15:10 PT (UTC -7)

Location: Merrill Hall

Session Type: Poster

Track: Adaptive Systems, Machine Learning, and Data Analytics

TP8a.1: Improved Multiplayer Bandit Algorithms for Bernoulli Rewards

Khang Nguyen, USC, United States; Ricardo Parada, William Chang, University of California, Riverside, United States

TP8a.2: Depth- and Rank-Selective Multilinear Transformation Layers for Compact CNNs

Van Tien Nguyen, The University of Texas at San Antonio, United States; Mayur Dhanaraj, Amazon, United States; Panagiotis Markopoulos, The University of Texas at San Antonio, United States

TP8a.3: ACTIVE SPEAKER DETECTION IN CROWDED COLLABORATIVE CLASSROOM ENVIRONMENTS

Sebastian Janampa, Marios Pattichis, The University of New Mexico, United States; Sylvia Celedon-Pattichis, The University of Texas at Austin, United States; Wenjing Shi, The University of New Mexico, United States

TP8a.4: CrossLag: Predicting Major Dengue Outbreak Signals with Limited Data and Transformers

Ashwin Prabu, Nhat Thanh Tran, Jack Xin, Guofa Zhou, University of California, Irvine, United States

TP8a.5: Entropy-aware Fine-grained Contextual Fusion for Multiple Autonomous Driving Conditions

Markus Flierl, Zhenghao Li, KTH Royal Institute of Technology, Sweden

TP8a.6: Towards Universal Representation-Based Process Control

Jinmyeong Choi, Taesup Kim, Seoul National University, Republic of Korea; Artur Dubrawski, Carnegie Mellon University, United States

TP8a.7: Unboxing the CNN: Mechanistic Interpretability and Inhibitory Logic in Time-Frequency Signal Classification

Thomas Kidd, Qi Cheng, Oklahoma State University, United States

TP8a.8: Robust Multi-Agent Bandits with Heavy-Tailed Rewards and Information Asymmetry

Ricardo Parada, William Chang, University of California, Riverside, United States

TP8a.9: Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network

Ziqing Li, University of Iowa, United States; Myung Cho, California State University, Northridge, United States; Qiutong Jin, University of California, Berkeley, United States; Weiyu Xu, University of Iowa, United States

TP8a.10: Can Rich Embeddings Compensate for Model Simplicity? Exploring the Relationship Between Embedding Quality and Model Complexity

Cate Dunham, Worcester Polytechnic Institute, United States; Thomas Cao, DTRA RD-CBEI A&AS, United States; Chia-Wei Tsai, DTRA RD-CBEI S&T Manager, United States; Joshua Uzarski, DEVCOM Soldier Center, United States; Randy Paffenroth, Worcester Polytechnic Institute, United States

TP8a.1: Improved Multiplayer Bandit Algorithms for Bernoulli Rewards

TP8a.2: Depth- and Rank-Selective Multilinear Transformation Layers for Compact CNNs

TP8a.3: ACTIVE SPEAKER DETECTION IN CROWDED COLLABORATIVE CLASSROOM ENVIRONMENTS

TP8a.4: CrossLag: Predicting Major Dengue Outbreak Signals with Limited Data and Transformers

TP8a.5: Entropy-aware Fine-grained Contextual Fusion for Multiple Autonomous Driving Conditions

TP8a.6: Towards Universal Representation-Based Process Control

TP8a.7: Unboxing the CNN: Mechanistic Interpretability and Inhibitory Logic in Time-Frequency Signal Classification

TP8a.8: Robust Multi-Agent Bandits with Heavy-Tailed Rewards and Information Asymmetry

TP8a.9: Repair Brain Damage: Real-Numbered Error Correction Code for Neural Network

TP8a.10: Can Rich Embeddings Compensate for Model Simplicity? Exploring the Relationship Between Embedding Quality and Model Complexity

TP8a.11: Coordinating the Unknown Lipschitz Constant in Multiplayer Bandits

TP8a.12: Ambiguity-Reduced Margin-Based Knowledge Distillation for Improved Calibration in Low-Capacity Models

TP8a.13: Synthetic Pretraining for Neural UAV Trajectory Prediction from Direction-Only Measurements

TP8a.14: Hilbert-Scale Stability of Posterior Diffusion Priors for Ill-Posed Signal Recovery

TP8a.15: When is Quantized Training Effective? A Data-Aware Monitoring Approach