All Articles
17939 articles total
Learner-based Concept Drift Detection: Analysis and Evaluation
A theoretical analysis and evaluation of various learner-based concept drift detection algorithms for streaming data environments.
ScholarQuest: A Taxonomy-Guided Benchmark for Agentic Academic Paper Search in Open Literature Environments
ScholarQuest, a new taxonomy-guided benchmark for evaluating LLM-based agents performing academic paper searches in open literature.
SPOT-E: Test-Time Entropy Shaping with Visual Spotlights for Frozen VLMs
SPOT-E is a test-time entropy shaping method that uses visual spotlights to improve the grounding and robustness of frozen Vision-Language Models.
Finetuning Vision-Language-Action Models Requires Fewer Layers Than You Think
Research showing that Vision-Language-Action (VLA) models can be compressed by up to 50% in depth without losing performance through a training-free structural compression pipeline.
The Register Gap: A Meaning Intelligence Framework for Nigerian Public Discourse
The Meaning Intelligence Framework (MIF) is proposed to better analyze Nigerian public discourse by separating surface sentiment from true communicative intent.
Editorial Alignment: A Participatory Approach to Engaging Editorial Expertise in LLM-mediated Knowledge Dissemination
A study on 'editorial alignment' as a design practice to ensure LLM-mediated knowledge dissemination adheres to the standards of public knowledge institutions.
ELVA: Exploring Ranking-Driven Universal Multimodal Retrieval
ELVA, a rule-based RL framework designed to mitigate 'grain blindness' in universal multimodal retrieval using MLLMs.
Boundary Embedding Shaping with Adaptive Contrastive Learning for Graph Structural Disentanglement
Boundary Embedding Shaping (BES) is introduced as an adaptive contrastive learning plug-in for GNNs to reduce structural noise at decision boundaries.
Giant Banana Pulled over in Montana Driver Says Cops Have Stopped Him 100s Times
A news story about a giant banana statue in Montana causing frequent police stops.
NASA selects Eric Schmidt’s rocket company for a 2028 mission to Mars
NASA has selected Eric Schmidt's rocket company, Relativity Space, to launch a Mars payload in 2028 to study the Martian atmosphere.
Sensorimotor World Models: Perception for Action via Inverse Dynamics
Introduction of Sensorimotor World Models (SMWM), a latent world model trained with inverse dynamics regularization to prevent representation collapse and align representations with actions.
Dual-Agent Framework for Cross-Model Verified Translation of Natural-Language Protocols into Robotic Laboratory Platform
A dual-agent framework using a Parser Agent and LLM Validation Agent to translate natural-language biological protocols into executable robotic lab commands.
Frequency-Aware Flow Matching for Continuous and Consistent Robotic Action Generation
Frequency-Aware Flow Matching (FAFM) is proposed for continuous and consistent robotic action generation using DCT and Sobolev-type constraints.
Hybrid ANN-SNN Pipeline with Local Plasticity
A hybrid ANN-SNN pipeline combining EfficientNet with a CoLaNET spiking classifier using local plasticity learning rules.
From Texts to Scores: Tracing the Emergence of Essay Quality Representations in Large Language Models
Research analyzing how LLMs encode essay quality representations linearly and identify specific 'essay scoring neurons'.
MedRLM: Recursive Multimodal Health Intelligence for Long-Context Clinical Reasoning, Sensor-Guided Screening, Evidence-Grounded Decision Support, and Community-to-Tertiary Referral Optimization
MedRLM is introduced as a recursive multimodal framework for long-context clinical reasoning and health intelligence using specialized agents and Evidence Graph Memory.
Evaluating and Enhancing Negation Comprehension in Remote Sensing MLLMs
The RS-Neg benchmark evaluates negation comprehension in remote sensing MLLMs, and the NeFo test-time learning method is proposed to improve it.
HilDA: Hierarchical Distillation with Diffusion for Advancing Self-Supervised LiDAR Pre-trainin
HilDA is a self-supervised pretraining framework for LiDAR backbones using hierarchical distillation and temporal occupancy diffusion.
I used sound waves to make espresso. It could cut coffee‑brewing energy use by ¾
An experimental project using sound waves to brew espresso, potentially reducing energy consumption by 75%.
A Neuromorphic Reinforcement Learning Framework for Efficient Pathfinding in Robotic Mobile Fulfillment Systems
Introduction of SDQN-RMFS, a framework that deploys RL-trained policies onto neuromorphic chips for ultra-low-power robotic pathfinding.