Apply deep learning to artificial intelligence and reinforcement learning using evolution strategies, A2C, and DDPG
InstructorLazy Programmer Inc.
The curriculum provides detailed coverage of A2C and DDPG algorithms, including implementation details for video environments.
Instruction emphasizes building methods from scratch rather than relying solely on high-level library abstractions.
Editorial course preview
These 4 complementary views highlight concrete, legible examples from the course presentation.
This diagram illustrates the core Reinforcement Learning loop, showing how an agent uses a Deep Neural Network to process states and take actions within an environment to maximize rewards.
This slide illustrates recent advances in robotics by displaying a grid of images showing a quadruped robot learning to walk in the real world through simulation training without hardware.
This slide illustrates the complexity of hand dexterity in robotics by showing a mechanical gripper manipulating a block, contrasting it with coarse locomotion.
This slide highlights recent AI advances in the game of Go, specifically noting how AlphaZero mastered the game using only self-play and referencing previous victories by AlphaGo.
The curriculum includes instructions for legacy libraries like Theano and contains specific lectures comparing Python 2 and 3, which may indicate outdated code requirements.
One signal suggests that code may fail due to upgraded versions of various packages in modern environments; however, this report predates the displayed update date and the update label does not prove a correction.
The curriculum aligns with the target of learning RL algorithms, though prerequisite depth varies.