↓Skip to main content

🏢 National Taiwan University

Trap-MID: Trapdoor-based Defense against Model Inversion Attacks
·3599 words·17 mins· loading · loading
AI Generated AI Theory Privacy 🏢 National Taiwan University
Trap-MID: Outsmarting model inversion attacks with cleverly placed ’trapdoors'!
TableRAG: Million-Token Table Understanding with Language Models
·2446 words·12 mins· loading · loading
Natural Language Processing Question Answering 🏢 National Taiwan University
TableRAG, a novel Retrieval-Augmented Generation framework, achieves state-of-the-art performance in large-scale table understanding by efficiently integrating schema and cell retrieval with language …
Robust Sparse Regression with Non-Isotropic Designs
·239 words·2 mins· loading · loading
AI Theory Robustness 🏢 National Taiwan University
New algorithms achieve near-optimal error rates for sparse linear regression, even under adversarial data corruption and heavy-tailed noise distributions.
Regularized Adaptive Momentum Dual Averaging with an Efficient Inexact Subproblem Solver for Training Structured Neural Network
·2322 words·11 mins· loading · loading
Machine Learning Deep Learning 🏢 National Taiwan University
RAMDA: a new algorithm ensures efficient training of structured neural networks by achieving optimal structure and outstanding predictive performance.
REBORN: Reinforcement-Learned Boundary Segmentation with Iterative Training for Unsupervised ASR
·2781 words·14 mins· loading · loading
AI Generated Natural Language Processing Speech Recognition 🏢 National Taiwan University
REBORN: An iterative training framework significantly improves unsupervised ASR by learning optimal speech segment boundaries using reinforcement learning, outperforming existing methods.
Hierarchical Programmatic Option Framework
·5774 words·28 mins· loading · loading
AI Generated Machine Learning Reinforcement Learning 🏢 National Taiwan University
Hierarchical Programmatic Option framework (HIPO) uses human-readable programs as options in reinforcement learning to solve long, repetitive tasks with improved interpretability and generalization.
Going Beyond Heuristics by Imposing Policy Improvement as a Constraint
·3576 words·17 mins· loading · loading
Machine Learning Reinforcement Learning 🏢 National Taiwan University
HEPO, a novel constrained optimization method, consistently surpasses heuristic-trained policies in reinforcement learning by ensuring policy improvement over heuristics, regardless of heuristic quali…