Supervised learning: The machine learning task of inferring a function from labeled training data.
Unsupervised learning: Solving various problems in pattern recognition based on training samples with unknown categories (not marked), which is called unsupervised learning.
Reinforcement learning: It is used to describe and solve the problem that the agent (agent) learns strategies to maximize rewards or achieve specific goals in the process of interacting with the environment.
Semi-supervised learning: Semi-supervised learning uses large amounts of unlabeled data, as well as labeled data, for pattern recognition.
Advantages: end-to-end problem solving, reducing the work of designing feature extractors for each problem, automatically extracting features, and being able to solve more complex tasks. By making full use of big data , the upper limit of accuracy is higher . But the theory is not complete.
Softmax + CE cross entropy loss derivation:
- LSTM: input gate, output gate, forget gate
Keep in mind: Wh+Ux
- GRU: reset gate, update gate
Remember: ReLU(AHW)