DiffRed: Dimensionality Reduction guided by stable rank
DiffRed introduces a novel approach to dimensionality reduction guided by stable rank, achieving lower distortion metrics compared to traditional techniques.
Sequence-Level Certainty Reduces Hallucination in Knowledge-Grounded Dialogue Generation
Sequence-level certainty reduces hallucination in Knowledge Grounded Dialogue Generation by proposing Certainty-based Response Ranking (CRR) methods.
SCOD: From Heuristics to Theory
Optimal SCOD strategy involves Bayes classifier for ID data and a linear selector in 2D space.
FERGI: Automatic Annotation of User Preferences for Text-to-Image Generation from Spontaneous Facial Expression Reaction
Facial expression analysis can automate user preference annotation for image generation tasks, improving scalability and efficiency.
InternVideo2: Scaling Video Foundation Models for Multimodal Video Understanding
InternVideo2 introduces a new video foundation model that excels in action recognition, video-text tasks, and video-centric dialogue through a progressive training paradigm.
DreamReward: Text-to-3D Generation with Human Preference
DreamReward proposes a comprehensive framework to improve text-to-3D models by learning from human preference feedback, resulting in high-fidelity and aligned 3D results.
SDP Synthesis of Maximum Coverage Trees for Probabilistic Planning under Control Constraints
The paper introduces MAXCOVAR BRT, a multi-query algorithm for probabilistic planning under control constraints with explicit coverage guarantees.
Conformal Monte Carlo Meta-Learners for Predictive Inference of Individual Treatment Effects
The Conformal Monte Carlo framework provides predictive distributions for individual treatment effects, aiding decision-making.
RMDPs: Average and Blackwell Optimality Study
RMDPsにおける平均とブラックウェル最適性の研究を探求する。
DREAMLLM: Synergistic Multimodal Comprehension and Creation at ICLR 2024
DREAMLLM introduces a learning framework that enhances multimodal comprehension and creation, showcasing superior performance in zero-shot tasks.