All Articles
17040 articles total
OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes
OmniPM-Net is a fusion model using Convolutional Conditional Neural Processes to reconcile discrete station-scale and gridded PM10 air quality forecasts.
SeqGPT: A Constrained Transformer Agent for the Inverse Designof Multi-Panel Composite Structures
SeqGPT is a constrained Transformer agent using neurosymbolic decoding to optimize composite stacking sequences for structural design.
Towards Self-Evolving Agents: A Human-Inspired Adaptive Exploration-Exploitation Framework for Genetic Network Programming
Introduction of HGNP, a human-inspired adaptive exploration-exploitation framework for Genetic Network Programming to evolve agentic AI strategies.
LinkedIn is a cesspool of scammers and identity theft
A discussion thread on Hacker News regarding the prevalence of scams and identity theft on LinkedIn.
FormalAnalyticGeo: A Neural-Symbolic Based Framework for Multimodal Analytic Geometry Problem Generation
Introduction of FormalAnalyticGeo, a framework for automatically generating multimodal analytic geometry problems and the accompanying AnalyticGeo7K dataset.
Resist and Update: Counterfactual Report Coordinates for Incentive-Compatible LLMs
A research paper proposing a method to make LLMs more incentive-compatible by learning counterfactual report mediators to resist user pressure.
Win by Silence: Deletion Non-Monotonicity, Autonomous Exploitation, and Typed-State Gating in LLM Plan Evaluation
An analysis of 'omission incentives' in LLM plan evaluation, where evaluators reward plans for being less explicit, and the introduction of PCSC to neutralize this.
Dynamic Resource Allocation for Ensemble Determinization MCTS
Proposed enhancements for Ensemble Determinization MCTS using dynamic resource allocation for better performance in adversarial board games.
Audio-Native Speech Recognition with a Frozen Discrete-Diffusion Language Model
Development of an audio-native speech recognition system using a frozen discrete-diffusion language model (DiffusionGemma) and a Whisper encoder.
Do AI Agents Know When a Task Is Simple? Toward Complexity-Aware Reasoning and Execution
The E3 (Estimate, Execute, Expand) framework aims to reduce LLM agent redundancy by estimating task complexity before execution.
Answering Without Referring: How AI Search Rewrites the Web's Economic Bargain
Research on how AI search engines reduce outbound clicks to the web, potentially disrupting the economic model of content production.
FAIR GraphRAG: A Retrieval-Augmented Generation Approach for Semantic Data Analysis
FAIR GraphRAG is introduced to integrate FAIR Digital Objects into graph-based retrieval systems for improved scientific data analysis.
Scaling Point-in-Time Language Models
Research on scaling Point-in-Time language models to eliminate lookahead bias for use in finance and social sciences.
Tracing Agentic Failure from the Flow of Success
Researchers propose OAT, an unsupervised failure attribution model for LLM agents that uses neural controlled differential equations to identify error steps without costly step-level annotations.
Accuracy and Normalized Accuracy under Length Bias: Analysis, Guidelines, and a Bayesian Alternative
This paper introduces Bayesian accuracy, a scoring rule that removes length bias in multiple-choice LLM benchmarks by computing posterior probabilities under an explicit length prior.
Do We Really Need Multimodal Emotion Language Models Larger Than 1B Parameters?
The paper introduces Light-MER, a lightweight multimodal emotion recognition framework that uses knowledge distillation to achieve high performance in sub-billion-parameter models.
Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents
The Double Ratchet framework proposes a method for the co-evolution of evaluation metrics and agent skills, allowing self-improving agents to function without pre-existing reliable metrics.
Visual Access Boundaries in Vision-Language Model Reasoning
Researchers analyze the Visual Access Boundary (VAB) in Vision-Language Models, finding that Chain-of-Thought reasoning relies more on language-side computation than continued image-token access.
Human-AI Agent Interaction as a Neuroplastic Training Environment
This research explores the neuroplastic effects of frequent AI agent interaction, suggesting a framework to mitigate negative reactive patterns through mindful observation.
Solution of the Hempel's statistical ambiguity problem and Causal AI
The paper addresses Hempel's statistical ambiguity problem by introducing Causal Rules and a semantic probabilistic inference procedure to achieve maximally specific causal relationships.