Dota 2 with Large Scale Deep Reinforcement Learning
No AI system could defeat world champions at Dota 2, an esports game with long time horizons, imperfect information and complex continuous state-action spaces.
Papers
1,138 of 1,138 verified cases.
Scientific Domain Knowledge Improves Exoplanet Transit Classification with Deep Learning
Physics & Astronomy
Stellar, planetary, and galactic studies
ai assisted
importance 6/10
scale stated
Space missions like Kepler and TESS produce large datasets of transit signals that must be efficiently and systematically classified as exoplanets or false positives.
Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning
Extracting information from millions of camera-trap wildlife images is expensive, time-consuming manual work.
Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks
The functional mechanisms of most disease-associated noncoding genetic variants are unknown, and the regulatory code of gene expression is incompletely understood.
Machine Learning-Guided Optimization of p -Coumaric Acid Production in Yeast
Biology & Life Sciences
Microbial Metabolic Engineering and Bioproduction
solved
importance 6/10
scale stated
Optimizing microbial cell factories (p-coumaric acid production in yeast) through DBTL cycles is slow without effective links between phases.
Deep learning the cis-regulatory code for gene expression in selected model plants
Biology & Life Sciences
Genomics and Chromatin Dynamics
partially solved
importance 6/10
scale stated
Understanding how non-coding regulatory DNA sequences determine gene expression levels in four plant species, including linking genetic variation to expression changes.
Deep structured learning for variant prioritization in Mendelian diseases
Identifying the disease-causing genetic variant among many candidates in Mendelian diseases is slow and difficult.
Mathematical discoveries from program search with large language models
LLM confabulations hinder scientific discovery, and the cap set problem in extremal combinatorics lacked constructions beyond the best-known ones.
Fast and scalable search of whole-slide images via self-supervised deep learning
Searching large repositories of gigapixel whole-slide images for morphologically similar slides was too slow because retrieval speed scaled with repository size.
Swarm learning for decentralized artificial intelligence in cancer histopathology
Training robust AI on medical images needs large multicentric datasets that face practical, ethical and legal obstacles to sharing.
Automated machine learning for endemic active tuberculosis prediction from multiplex serological data
Active tuberculosis serological diagnosis requires interpreting complex multi-antibody data across variable immune responses, and building suitable algorithms manually is slow.
Accurate diagnosis of lymphoma on whole-slide histopathology images using deep learning
Histopathological diagnosis of lymphomas, especially follicular lymphoma, is challenging, requires expertise or centralized review, and depends on tissue-section technical quality.
Deep learning enables structured illumination microscopy with low light levels and enhanced speed
Biology & Life Sciences
Advanced Fluorescence Microscopy Techniques
solved
importance 6/10
scale stated
Structured illumination microscopy needs intense illumination and many acquisitions per image, causing photobleaching and limiting live-cell super-resolution imaging.
Predicting day-ahead solar irradiance through gated recurrent unit using weather forecasting data
Solar power operators need accurate day-ahead irradiance forecasts for planning, scheduling and managing photovoltaic plants and grid-connected systems.
Deep learning enhanced light sheet fluorescence microscopy for in vivo 4D imaging of zebrafish heart beating
Biology & Life Sciences
Advanced Fluorescence Microscopy Techniques
solved
importance 6/10
scale stated
Time-resolved volumetric fluorescence imaging over extended durations is limited by the trade-off among imaging speed, light exposure, illumination power and image quality.
Validating the accuracy of deep learning for the diagnosis of pneumonia on chest x-ray against a robust multimodal reference diagnosis: a post hoc analysis of two prospective studies
Deep learning pneumonia diagnosis on chest X-rays had mainly been validated against radiologists, whose own diagnoses can be inaccurate, so accuracy was tested against a robust multimodal expert reference.
Interpretable machine learning for materials discovery: Predicting CO2 adsorption properties of metal–organic frameworks
Engineering & Materials
Machine Learning in Materials Science
ai assisted
importance 6/10
scale stated
Screening millions of hypothetical metal–organic frameworks to find ones with high CO2 adsorption capacity is computationally expensive.
Machine Learning as a Tool for Hypothesis Generation
Research & Humanities
Explainable Artificial Intelligence (XAI)
ai assisted
importance 6/10
scale stated
Hypothesis generation in social science remains informal; specifically, understanding what facial features drive judges' jailing decisions.
Deep model predictive control of gene expression in thousands of single cells
Researchers could only passively observe natural gene expression dynamics in cells; they needed a way to actively impose specific, cell-specific expression dynamics in individual live cells.
Polymer design via SHAP and Bayesian machine learning optimizes pDNA and CRISPR ribonucleoprotein delivery
Designing polymer carriers that efficiently deliver pDNA and CRISPR-Cas9 ribonucleoprotein into cells is hard because the structure-property relationships in combinatorial polymer libraries are complex and poorly understood.
Deep learning enhancing guide RNA design for CRISPR/Cas12a‐based diagnostics
Designing effective crRNA guides for Cas12a-based nucleic acid detection is challenging and time-consuming, slowing development of rapid pathogen tests.
Accurate Prediction of Protein Structural Flexibility by Deep Learning Integrating Intricate Atomic Structures and Cryo-EM Density Information
Biology & Life Sciences
Protein Structure and Dynamics
partially solved
importance 6/10
scale stated
Computationally predicting protein structural flexibility/dynamics was challenging and existing methods were inaccurate.
Prediction of base editor off-targets by deep learning
Biology & Life Sciences
CRISPR and Genetic Engineering
partially solved
importance 6/10
scale stated
Base editors frequently induce unwanted Cas9-dependent off-target DNA mutations, and researchers lacked tools to predict where these off-targets occur.
Deep learning integrates histopathology and proteogenomics at a pan-cancer level
Predicting clinical outcomes and molecular features (e.g., TP53 status) from routine H&E pathology slides alone, and explaining which histology features are predictive, was difficult.