David Ha is an AI researcher known for pioneering work in neuroevolution, reinforcement learning, generative models, and world models. His blog features influential experiments, tutorials, and creative AI projects blending deep learning with evolutionary algorithms.
9 articles from this blog
A survey exploring how concepts from collective intelligence, like swarm behavior and emergence, are being applied to improve deep learning systems.
EvoJAX is a hardware-accelerated neuroevolution toolkit built on JAX for running parallel evolution experiments on TPUs/GPUs.
Introduces permutation-invariant neural networks for RL agents, enabling robustness to shuffled, noisy, or incomplete sensory inputs.
Introduces PlaNet, a model-based AI agent that learns environment dynamics from pixels and plans actions in latent space for efficient control tasks.
Step-by-step guide to reproducing the 'World Models' AI experiments, including prerequisites, software setup, and instructions for running pre-trained models.
Explores applying Evolution Strategies (ES) to reinforcement learning problems for finding stable and robust neural network policies.
A visual guide explaining Evolution Strategies (ES) as a gradient-free optimization alternative to reinforcement learning for training neural networks.
Explores a neural network model, sketch-rnn, that generates vector drawings by learning from human sketch sequences, mimicking abstract visual concepts.
A tutorial for artists on using a pre-trained recurrent neural network with Javascript and p5.js to generate interactive handwriting and vector artwork.
