17
XFACTORS: Disentangled Information Bottleneck via Contrastive Supervision
The XFACTORS framework combines contrastive supervision with the information bottleneck principle to achieve stable and scalable disentangled representation learning.
Top stories in Research
17
MAVEN: A Multi-stage Agentic Annotation Pipeline for Video Reasoning Tasks
Researchers have introduced MAVEN, a multi-stage agentic pipeline designed to generate high-quality structured video annotations and reasoning traces for training vision-language models.
17
Feedback Manipulation Regularization: Enabling Offline Agent Alignment for Imitation Learning
Researchers have introduced a regularization method to improve offline agent alignment in imitation learning by leveraging human feedback and demonstrations.
16
Architecture Generalization with MetaNCA
The MetaNCA framework is introduced to improve the adaptability and architectural generalization of neural cellular automata models through localized interactions.
16
OmniFood-Bench: Evaluating VLMs for Nutrient Reasoning and Personalized Health Advice
OmniFood-Bench is a new benchmark designed to assess how well vision-language models can reason about nutritional content and provide personalized dietary advice.
16
Spectral Analysis of Dueling Q-Learning
This paper provides a theoretical spectral analysis of dueling Q-learning to better understand its convergence properties and performance in reinforcement learning.
16
MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection
MSRNet is a multi-scale recursive network designed to improve the detection and segmentation of camouflaged objects in challenging visual environments.
16
MetaHGNIE: Meta-Path Induced Hypergraph Contrastive Learning in Heterogeneous Knowledge Graphs
The proposed MetaHGNIE framework utilizes hypergraph contrastive learning to better capture high-order interactions and estimate node importance in heterogeneous knowledge graphs.
16
SwinIFS: Landmark Guided Swin Transformer For Identity Preserving Face Super Resolution
SwinIFS utilizes landmark guidance and hierarchical attention mechanisms within a Swin Transformer to reconstruct high-quality facial images while preserving identity details.
16
AI-guided stimuli discovery and generation to optimize facial emotion perception studies in autism
This study utilizes generative AI to create precise facial emotion stimuli, helping researchers better analyze perceptual differences in autistic individuals.
16
How novice coders can develop AI programs for military applications
Researchers from MIT Lincoln Laboratory and the US Air Force found that chatbots can enable non-technical military personnel to build functional software applications.
15
Huawei’s next smartphone chip taps new scaling law for performance boost: paper
Huawei has unveiled its LogicFolding architecture for the upcoming Kirin 2026 mobile processor, which reportedly increases transistor density by 55 percent to boost performance without relying on more advanced lithography nodes.
15
Beyond Thermal Imaging: Inferring Thermophysical Properties from Time-Resolved Thermal Observations
Researchers have developed a method to infer latent thermophysical properties of a scene from time-resolved thermal observations.
15
SHAP-Weighted Cross-Modal Expert Fusion for Emotion and Sentiment Recognition: Evidence and Limits
The paper evaluates an explainable AI-guided adaptive fusion method that combines unimodal and cross-modal experts for multimodal emotion and sentiment recognition.
15
Goal-Driven Reasoning in DatalogMTL with Magic Sets
A new goal-driven reasoning method for DatalogMTL utilizes the magic sets technique to improve computational efficiency in temporal reasoning tasks.
15
ASMR: Agentic Schema Generation for Ship Maintenance Report Writing
Researchers have developed ASMR, an agentic framework designed to automatically generate structured schemas from historical ship maintenance reports.
15
PS4: Proxy-Supervised Joint Training for Real Target Speaker Extraction
The PS4 framework uses proxy supervision and a newly compiled dataset to train target speaker extraction models on real conversational audio without clean reference signals.
15
Spatio-Temporal Scheduling Prediction Under Backhaul Delay for Resilient Coordinated Beamforming
This paper addresses backhaul latency in 5G networks by using predictive scheduling to maintain the performance of coordinated beamforming.
15
MobiDiff: Semantic-Aware Multi-Channel Discrete Diffusion for Human Mobility Data Generation
The MobiDiff framework uses discrete diffusion to generate realistic human mobility data while preserving privacy and modeling semantic events.
15
Drift-Aware Temporal Graph Rewiring (DATGR) for Adaptive Semantic Modeling in Biomedical Text
The Drift-Aware Temporal Graph Rewiring framework dynamically updates concept relationships to help language models adapt to evolving terminology in biomedical literature.
14
Evaluating the Effect of Frame Rate in Sequence-Based Classification of Autism-Related Self-Stimulatory Hand Idiosyncrasies
This study evaluates how temporal sampling rates and neural network architectures affect the automated detection of autism-related self-stimulatory behaviors in video.
14
Agentic AI and Retrieval-Augmented Models in Straight-Through Underwriting
This paper explores the integration of agentic AI and retrieval-augmented generation models to automate complex decision workflows in actuarial underwriting.
14
A Graph Neural Network Model for Real-Time Gesture Recognition Based on sEMG Signals
Researchers developed a graph neural network model that analyzes muscle activation patterns to achieve real-time hand gesture recognition for prosthetics and augmented reality.
14
INTENT: An LSTM Framework for Vehicle Intention Prediction in Intersection Scenarios with Comprehensive Ablation Analysis
Researchers developed an LSTM-based framework designed to predict vehicle intentions at complex road intersections to improve autonomous driving safety.