Reinforcement Learning with Augmented Data
Explore how data augmentation enhances reinforcement learning algorithms, improving efficiency and generalization across visual observation tasks without modifying core RL methods.
Chip Placement with Deep Reinforcement Learning
Deep reinforcement learning optimizes chip placement, outperforming human experts in speed and efficiency. This AI-driven approach learns from experience, generalizes to new designs, and minimizes power, performance, and area constraints.
The AI Economist - Improving Equality and Productivity with AI-Driven Tax Policies
Explore AI-driven tax policies that balance equality and productivity through reinforcement learning, revealing emergent strategies and outperforming traditional economic models in simulations and human experiments.
Thinking While Moving - Deep Reinforcement Learning with Concurrent Control
Explore concurrent control in reinforcement learning, where agents make decisions while executing actions, revolutionizing robotic movement and decision-making processes.
Gradient Surgery for Multi-Task Learning
Explore PCGrad, a technique addressing gradient conflicts in multi-task learning, enhancing efficiency and performance in supervised and reinforcement learning scenarios.
Dynamical Distance Learning for Semi-Supervised and Unsupervised Skill Discovery
Explore dynamical distance learning for reinforcement agents, enhancing policy learning through state distance estimation. Discover applications in semi-supervised and unsupervised skill acquisition.
CURL - Contrastive Unsupervised Representations for Reinforcement Learning
Explore how contrastive learning enhances reinforcement learning, improving performance in complex tasks using high-level features extracted from raw pixels.
Dream to Control - Learning Behaviors by Latent Imagination
Explore DeepMind's Dreamer, an innovative RL agent that learns complex behaviors through latent imagination, surpassing existing approaches in visual control tasks.
Agent57 - Outperforming the Atari Human Benchmark
Explore DeepMind's Agent57, the first AI to surpass human performance in all 57 Atari games. Learn about its innovative approach to reinforcement learning and meta-learning for exploration-exploitation balance.
A Neurally Plausible Model Learns Successor Representations in Partially Observable Environments
Explores a neurally plausible model using distributional successor features for efficient reinforcement learning in partially observable, noisy environments, bridging model-based and model-free approaches.
AlphaStar - Grandmaster Level in StarCraft II Using Multi-Agent Reinforcement Learning
Explore DeepMind's AI agent achieving Grandmaster-level skill in StarCraft II through innovative multi-agent reinforcement learning and League Training techniques.
IMPALA - Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
Explore IMPALA, a distributed reinforcement learning architecture for efficient multi-task learning. Learn about V-trace, off-policy correction, and scalable performance across thousands of machines.
Reinforcement Learning, Fast and Slow
Explores recent advancements in deep reinforcement learning, discussing techniques for faster learning and their potential implications for cognitive science and neuroscience.
Datasets for Data-Driven Reinforcement Learning
Explore offline reinforcement learning benchmarks and datasets for evaluating algorithms in various settings, including safety-critical tasks and learning from expert demonstrations.
Go-Explore: A New Approach for Hard-Exploration Problems
Discover a groundbreaking algorithm that outperforms in hard-exploration Atari games, utilizing random exploration and memory to achieve superhuman scores in Montezuma's Revenge and Pitfall.
MuZero - Mastering Atari, Go, Chess, and Shogi by Planning with a Learned Model
MuZero: AI algorithm combining tree-based search with learned model for superhuman performance in complex domains without prior knowledge of dynamics. Predicts rewards, policies, and value functions.
Learning World Graphs to Accelerate Hierarchical Reinforcement Learning
Explore hierarchical reinforcement learning using world graphs to accelerate learning in complex environments. Learn about pivotal states, goal-conditioned policies, and efficient task solving.
World Models
Explore generative neural network models for reinforcement learning environments, learning compressed representations and training agents in hallucinated dreams for efficient task-solving.
Curiosity-Driven Exploration by Self-Supervised Prediction
Explore how curiosity can be an intrinsic reward for agents in sparse-reward environments, using self-supervised prediction to enable efficient exploration and skill acquisition.
Imagination-Augmented Agents for Deep Reinforcement Learning
Explore innovative deep reinforcement learning architecture combining model-free and model-based approaches for improved performance and efficiency in complex environments.
ChatGPT - This AI Has a Jailbreak! - Unbelievable AI Progress
Explore ChatGPT's capabilities, origins, and potential risks. Learn about its underlying technology, amazing features, and ways to bypass safety mechanisms, while considering OpenAI's vision for the future.