Science – Page 77

Can AI Lab Assistants Be Trusted?

15.03.2026 by ebaster

A new benchmark reveals significant gaps in the safety reasoning of large language models tasked with operating in complex scientific environments.

Navigating Complex Networks with Maximum-Entropy Random Walks

15.03.2026 by ebaster

$Directed hypergraphs facilitate the broadcasting of messages across complex networks, where activation flows from pivotal nodes to interconnected receiver sets-a mechanism illustrated by the propagation along edges such as [latex]v_1 \rightarrow \{v_2, v_3\} [/latex]-enabling communication pathways beyond traditional pairwise connections.$

A new framework leverages maximum-entropy principles to model and analyze dynamics on hypergraphs, offering insights into complex relationships beyond traditional network structures.

What Video AI ‘Sees’: Decoding Action and Outcome

15.03.2026 by ebaster

The study demonstrates that individual Multi-Layer Perceptron (MLP) blocks possess causal sufficiency in generating the layer 11 signal, indicating that focused intervention at these components is sufficient to influence the system's outcome without needing to consider broader network interactions - a testament to localized control within a complex architecture.

New research reveals how video AI models internally represent the nuances of actions, distinguishing between successful and unsuccessful attempts even when the final classification remains the same.

Beyond Devices: Self-Learning IMU Recognition for Universal Gesture Control

15.03.2026 by ebaster

UniMotion demonstrates robust cross-task generalization, achieving an average accuracy of 93% on three human activity recognition datasets-Shoaib, MotionSense, and UCI-after pre-training on the unlabeled HHAR dataset and fine-tuning with only 10% labeled data.

A new self-supervised learning framework unlocks accurate and adaptable motion recognition across a wide range of devices and users, minimizing the need for labeled data.

Untangling Complexity: A Machine Learning Approach to Symbolic Simplification

15.03.2026 by ebaster

The system leverages a Transformer encoder to process feature vectors of symbolic terms, intentionally omitting positional encoding to maintain permutation symmetry, and subsequently employs a permutation equivariant policy head to generate probabilities over possible simplification actions, effectively learning to navigate the action space without regard to term order.

Researchers have developed a self-supervised learning method that allows AI to learn how to simplify complex mathematical expressions by reversing the process of ‘scrambling’ them.

Orchestrating Care with Intelligent Agents

15.03.2026 by ebaster

A new system architecture aims to bring the power of large language models to dynamic clinical workflows with enhanced safety and coordination.

Beyond the Buzz: Measuring the Real Value of AI Code Review

15.03.2026 by ebaster

A new benchmark reveals that while AI code review tools can find many potential issues, prioritizing accuracy over sheer volume is crucial for effective defect detection.

Building Worlds with Words: A New Framework for AI Navigation

15.03.2026 by ebaster

The MANSION framework constructs multi-story 3D buildings from natural language descriptions through a sequential pipeline-beginning with whole building planning, refining to per-floor layouts, synthesizing floorplans, and ultimately instantiating the complete scene-demonstrating a system capable of translating linguistic intent into architectural form.

Researchers have developed a novel system that generates complex, multi-story 3D environments from natural language descriptions, enabling more realistic and challenging tests for embodied AI agents.

Decoding the Language of Genes: A New Deep Learning Approach

15.03.2026 by ebaster

Temporal Convolutional Networks, as formalized by Bai et al., offer a method for sequence modeling that leverages the inherent order of data without relying on recurrent architectures.

Researchers have developed a novel deep learning framework to more accurately predict how proteins bind to DNA and regulate gene expression.

Swarm Mapping: Collaborative Robots Navigate with Edge Computing

15.03.2026 by ebaster

A distributed system enables collaborative simultaneous localization and mapping (SLAM) across multiple robots through a dedicated communication architecture, allowing for shared environmental understanding and coordinated navigation.

A new system leverages cloud-edge collaboration and efficient data compression to enable accurate and robust multi-robot simultaneous localization and mapping.