Papers

3,241 in the directory.

Browse AI research papers and methods. Search paper profiles, read recorded summaries and follow source links for the full research.

A–Z Recently synced

When Do Supervised UQ Ensembles Improve LLM Hallucination Detection? A Robustness Study taaft Uncertainty quantification (UQ) methods are widely used for hallucination detection in large langua… 2026-08-25 published
When Does a Classifier Help an LLM? Classifier-Guided Prompting and Hybrid Classifier-LLM Models for Credit-Default Prediction taaft Credit-default prediction is an important task in financial decision making. Traditional methods us… 2026-08-30 published
When Does an Interpretation Count as Established? The Formation, Evaluation, and Responsibility of Interpretation in Generative AI taaft Generative AI research has increasingly evaluated factuality, citation, coverage, and report struct… 2026-09-04 published
When Does Content-Based Routing Work? Representation Requirements for Selective Attention in Hybrid Sequence Models taaft We identify a routing paradox in hybrid sequence models: content-based routing - deciding which tok… 2026-03-22 published
When Does Information Sharing Improve Decentralized Discovery? Aggregation, Independent Rescue, and Equilibrium Selection taaft Information sharing can improve a pooled estimate while eliminating independent rescue actions. Thi… 2026-09-01 published
When Does LeJEPA Learn a World Model? taaft A representation that scrambles the true degrees of freedom of the world cannot support reliable pl… 2026-05-25 published
When Does Online Adaptation Pay on the Edge? A Leakage-Free Evaluation of Warmup, Learning-Rate Selection, and Resource Trade-offs for Time-Series Forecasting taaft Online adaptation can help edge time-series forecasting under distribution drift, but its measured… 2026-09-01 published
When Errors Become Memories: Causal Pathway Tracing in Multi-Turn Memory-Augmented LLMs taaft Long-term memory enables large language models (LLMs) to preserve and reuse information across inte… 2026-08-31 published
When Evidence Shapes Collaboration: Knowledge-Conditioned Topology Generation for Multi-Agent Systems taaft Multi-Agent Systems (MAS) have recently moved from static workflows toward dynamically generated co… 2026-08-28 published
When Financial Fine-tuning Fails: A Three-Level Detectability Analysis of Numerical Hallucination in Domain-Adapted Language Models taaft Financial large language models are increasingly deployed for summarization of reports and disclosu… 2026-09-04 published
When Genomic Masking Priors Fail to Transfer: Strong Variant Prediction, Weak Functional Generation taaft Bidirectional discrete diffusion model appears naturally suited to genomic modeling because it can… 2026-09-04 published
When Identities Collapse: A Stress-Test Benchmark for Multi-Subject Personalization taaft Subject-driven text-to-image diffusion models have achieved remarkable success in preserving single… 2026-03-27 published
When Language Models Lose Their Mind: The Consequences of Brain Misalignment taaft While brain-aligned large language models (LLMs) have garnered attention for their potential as cog… 2026-03-24 published
When Less is More: Investigating Data Pruning for Pretraining LLMs at Scale taaft Large volumes of text data have contributed significantly to the development of large language mode… 2023-09-08 published
When Less is More: Understanding When Token Filtering Helps and Fails in AI-generated Text Detection taaft The rapid advancement of large language models (LLMs) has made AI-generated text detection increasi… 2026-08-30 published
When Linguistic and Internal Confidence Diverge in Large Language Models taaft Users often ask large language models (LLMs) to report how confident they are, but it is unclear wh… 2026-08-28 published
When LLM Decompilers Recompile More and Preserve Less taaft Decompilation recovers high-level source from compiled machine code and serves as a foundation for… 2026-09-04 published
When LLM Meets Tree Search: A Systematic View of Inference as Search in Large Language Models taaft As pretraining scaling laws approach saturation, Test-Time Scaling (TTS) has emerged as an importan… 2026-08-31 published
When Models Edit Too Much: On the Fidelity of Minimal Code Edits taaft Large language models (LLMs) are increasingly used to edit existing code, but correctness alone is… 2026-09-03 published
When Models Hear What They Expect: Diagnosing Prosodic Heuristics in Multimodal Sarcasm Detection taaft Multimodal Large Language Models (MLLMs) process speech and text jointly, yet whether they exploit… 2026-08-31 published
When Models Judge Themselves: Unsupervised Self-Evolution for Multimodal Reasoning taaft Recent progress in multimodal large language models has led to strong performance on reasoning task… 2026-03-22 published
When Obstacles Bend: Modeling Vegetation Deformation in the context of Field Robotics taaft Autonomous robots operating in natural environments must often interact with vegetation rather than… 2026-08-26 published
When Optimization Becomes Manipulation: Defending Generative Search against Malicious Generative Engine Optimization taaft This paper focuses on defending generative search engines against malicious Generative Engine Optim… 2026-09-02 published
When Perplexity Lies: Generation-Focused Distillation of Hybrid Sequence Models taaft Converting a pretrained Transformer into a more efficient hybrid model through distillation offers… 2026-03-27 published
When Persona Attributes Improve Population Alignment in Large Language Models taaft Large Language Models (LLMs) are increasingly used to predict the responses of human participants i… 2026-09-02 published
When Personality Meets Quantization: A Layer-wise MBTI Analysis of Quantized LLMs taaft Personality is increasingly important in large language models (LLMs), as it shapes users' trust, e… 2026-08-26 published
When RAG Fails to Equalize: Geo-bias in Factual Question Answering over Public Companies taaft Retrieval-augmented generation (RAG) is widely assumed to mitigate factual errors in large language… 2026-08-26 published
When Retrieval Helps: Selective Retrieval for Single-Turn Mental-Health QA taaft Retrieval-augmented generation (RAG) can improve the specificity and grounding of large language mo… 2026-09-03 published
When RL Fails after SFT: Rejuvenating Model Plasticity for Robust SFT-to-RL Handoff taaft Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL) has become a standard pipeline… 2026-06-07 published
When Safety Routing Breaks: Understanding Alignment Fragility under Benign Fine-Tuning taaft Benign fine-tuning severely weakens the safety alignment of large language models (LLMs), so we stu… 2026-09-01 published
When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMs taaft Safety alignment of large language models (LLMs) degrades across languages, yet the internal mechan… 2026-08-30 published
When Should a Network Emit Geometry, and When Should It Detect It? Readout, Reconciliation, and Representation in Floorplan Vectorization taaft A network trained to recover the walls, openings, and rooms of a rasterized floorplan can produce i… 2026-08-26 published
When Similarity Is Interaction-Driven: Quantum Kernels for Regime-Sensitive Learning taaft Similarity in many decision systems is governed not by distance alone but by interactions among var… 2026-08-25 published
When Text Misleads: Inconsistent-Aware Reasoning for Audio-Grounded Dialogue taaft Understanding spoken dialogue requires joint reasoning over lexical content and paralinguistic acou… 2026-08-27 published
When the Algorithm Becomes the Brand Crisis: A Sociotechnical Theory of Distributed Responsibility and Accountable Transparency taaft Artificial intelligence systems increasingly enact market-facing promises through chatbots, recomme… 2026-09-01 published
When the Martingale Never Stops Firing: Anytime-Valid Gating on Real Forecast Streams taaft Machine learning systems are increasingly corrected while they run, and the decision of when to int… 2026-08-31 published