Deep Learning

Proximal Policy Optimization (PPO)

Proximal Policy Optimization (PPO)

7 min read Reinforcement Learning
In previous sections on Policy Gradient and REINFORCE, we introduced the core concepts of reinforcement learning (RL) algorithms that rely on gradients applied directly to the policy. These are known as policy-based approaches
Deep Q Network (DQN)

Deep Q Network (DQN)

5 min read Reinforcement Learning
DQN extends Q-Learning simply by replacing Q-table with a neural network, enabling it to handle high-dimensional and continuous state spaces, such as images in video games [Playing Atari with Deep Reinforcement Learning]. In
Transformer: Multi-Head Attention

Transformer: Multi-Head Attention

5 min read Deep Learning
Today, we’re going to dive deeper into the Transformer. However, before discussing its architecture, there's one important concept we need to cover: Multi-Head Attention. If you're not familiar
Transformer: Self-Attention

Transformer: Self-Attention

4 min read Transformer
Recently, one of my friends used LSTM with PPO to train a robot in a simulation aimed at solving a collection task. With a basic understanding of RNNs and LSTMs—an optimized form