PulseAugur / Brief
EN
LIVE 11:45:54

Brief

last 24h
[2/2] 224 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. RESEARCH · arXiv cs.CV English(EN) · · [2 sources]

    Self-supervised Learning Matters: A Simple Ensemble Solution for Micro-Gesture Recognition

    Researchers from XInsight Lab have developed a novel ensemble framework for micro-gesture recognition, achieving a new state-of-the-art result in the 4th MiGA Challenge at IJCAI 2026. Their approach integrates a self-supervised RGB model, pre-trained on a large unlabeled video dataset, with existing supervised models. This self-supervised component significantly improved performance, reaching 74.419% top-1 accuracy and outperforming previous benchmarks by over 1.2 percentage points. AI

    IMPACT Demonstrates the effectiveness of self-supervised learning for specialized visual recognition tasks, potentially improving performance in areas like human-computer interaction.

  2. TOOL · Mastodon — fosstodon.org English(EN) ·

    🚀 Excited to share that a survey paper from our RCLN team has been accepted at IJCAI 2026! This work has been done in collaboration with CentraleSupélec. "Graph

    A survey paper on graph rewiring techniques for Graph Neural Networks (GNNs) has been accepted at IJCAI 2026. The paper addresses the challenges of over-squashing and over-smoothing in GNNs, which hinder information flow and performance. It proposes graph rewiring as a method to improve GNNs by modifying their topology and opens a discussion on the necessity and attribution of improvements from such interventions. AI

    IMPACT This survey provides a structured overview of graph rewiring techniques, potentially guiding future research and development in GNNs for complex data structures.