PulseAugur / Brief
EN
LIVE 13:22:46

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.NE (Neural & Evolutionary) English(EN) · · [2 sources]

    Harnessing cortical geometry, wiring, and function as inductive biases for recurrent neural networks

    Researchers have developed biologically grounded recurrent neural networks by leveraging data from the MICrONS program, which combines electron microscopy and calcium imaging of mouse visual cortex. These networks utilize neuronal spatial coordinates, anatomical connectivity, and function-derived relationships from nearly 12,000 neurons to initialize weights and impose spatial constraints during learning. The study found that networks incorporating cortical structure and function significantly outperformed baseline models across three cognitive decision-making tasks, with functional weight initialization providing the most substantial gains. AI

    IMPACT Biologically inspired network architectures may lead to more efficient and effective learning algorithms.

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

    Instance-Aware Pseudo-Labeling and Class-Focused Contrastive Learning for Weakly Supervised Domain Adaptive Segmentation of Electron Microscopy

    Researchers have developed new methods for domain adaptive segmentation of electron microscopy images, crucial for biological and neuroscience research. The first approach, Instance-Aware Pseudo-Labeling and Class-Focused Contrastive Learning, uses sparse point labels and a multitask learning framework to improve segmentation accuracy. The second method, Prefer-DAS, introduces sparse promptable learning and local preference alignment, allowing for interactive segmentation and outperforming existing unsupervised and weakly-supervised techniques. AI

    Instance-Aware Pseudo-Labeling and Class-Focused Contrastive Learning for Weakly Supervised Domain Adaptive Segmentation of Electron Microscopy

    IMPACT These advancements in annotation-efficient segmentation could accelerate biological and neuroscience research by reducing the need for extensive manual labeling.