Science – Page 81

Mapping the Unknown: AI-Powered Swarms for Environmental Intelligence

01.02.2026 by ebaster

The study demonstrates that an AI-augmented Distributed Two-Objective Coverage (D2OC) approach significantly improves coverage efficiency-as evidenced by the refined sample distribution-compared to a standard mission completion without AI, a performance gain achieved through an adaptive learning process monitored by the decreasing trajectory of the loss function [latex] L [/latex].

A new decentralized control framework leverages artificial intelligence to enable multi-agent systems to efficiently map complex environments.

Seeing is Believing: Predicting Gene Expression Directly from Cancer Histology

01.02.2026 by ebaster

HistoPrism establishes an architecture wherein image embeddings derived from pathology foundation models are contextually refined through a Transformer Encoder and pan-cancer conditioning via cross-attention, ultimately predicting gene expression values with a final multi-layer perceptron head-a system designed not by construction, but by propagating relationships inherent in the data itself.

A new model accurately decodes biological activity from tissue images, offering a powerful way to analyze cancer pathways without relying on traditional genomic data.

Thinking Agents: Rewarding Reasoning for Smarter AI

31.01.2026 by ebaster

The study demonstrates that the weighting of the reward function in an Agent-based Reinforcement Learning system-specifically, the parameter λ associated with the Reward-Risk Metric-directly influences task performance.

New research explores how providing feedback specifically on an agent’s reasoning process can dramatically improve its ability to solve complex tasks and utilize tools effectively.

Unlocking Mobile App Coverage with AI

31.01.2026 by ebaster

CovAgent navigates interactive app elements within McDonald’s Canada-specifically those inaccessible to established automation frameworks like LLMDroid, Ape, and Fastbot-by leveraging dynamically instrumented edges, demonstrating an expanded capacity for complex app interaction.

A new framework leverages artificial intelligence to automatically explore and test previously unreachable code within Android applications.

Ask and You Shall Solve: Teaching AI to Question Its Way to Better Answers

31.01.2026 by ebaster

New research explores how large language models can move beyond simply answering questions to actively seeking clarification, dramatically improving their performance on complex reasoning tasks.

AI-Powered Design: Automating Domain Modeling with Generative AI

31.01.2026 by ebaster

Data preparation-encompassing cleaning, segmentation, and partitioning-forms the essential foundation upon which effective model training relies, transforming raw information into usable datasets.

New research demonstrates how generative artificial intelligence can significantly streamline the creation of domain-driven design models, offering a pathway to faster and more efficient software development.

Untangling Skill: Smarter Cloth Manipulation with Limited Data

31.01.2026 by ebaster

An agent is trained within a simulated environment with complete access to cloth state, then its knowledge is transferred to a real-world counterpart via a dataset generated and densely labeled by the simulation itself, effectively bridging the reality gap through self-supervised distillation.

Researchers have developed a new reinforcement learning approach that improves data efficiency in teaching robots to manipulate cloth, even without human demonstrations.

Smarter Agents: Streamlining AI Workflows for Peak Performance

31.01.2026 by ebaster

A new framework dramatically improves the efficiency of AI agents by consolidating repetitive tasks and reducing reliance on costly large language model calls.

Surgical Skill, Amplified: Learning from Expert Movements

31.01.2026 by ebaster

The study demonstrates a MoE-ACT policy’s adaptability during porcine surgery, successfully navigating distinct phases - reach, grasp, and retract - and suggesting a system where capability emerges from iterative refinement rather than predetermined design.

A new approach leverages the power of mixture-of-experts to significantly improve robotic surgery policies, even with limited training data.

Seeing Beyond the Slide: AI Predicts Prostate Cancer Recurrence

31.01.2026 by ebaster

A new artificial intelligence model accurately forecasts biochemical recurrence after prostatectomy by analyzing microscopic images of biopsy samples.