ADL2019

Applied Deep Learning (2019 Spring) @ NTU

This course is lectured by Yun-Nung (Vivian) Chen and has four homeworks. The four homeworks are as follows:

1. Dialogue Modeling
2. Contextual Embeddings
3. Deep Reinforcement Learning
4. Conditional Generative Adversarial Nets

Browse this course website for more details.

Table of Contents

1. Dialogue Modeling
   1. Data Preprocessing
   2. Training and Prediction
   3. Results (Recall@10)
2. Sequence Classification with Contextual Embeddings
   1. Part 1. Train an ELMo to beat the simple baseline
   2. Part 2. Beat the strong baseline with nearly no limitation
3. Deep Reinforcement Learning
   1. Policy Gradient
   2. Deep Q-Learning (DQN)
   3. Actor-Critic
4. Conditional Generative Adversarial Nets
   1. Cartoon Set
   2. Evaluation
   3. Train Condiction GANs
   4. Training Tips for Improvement
   5. Evaluate Condiction GANs
   6. FID Scores
   7. Training Progress
   8. Loss and Accuracy
   9. Human Evaluation Results

Results of Four Homeworks

1. Dialogue Modeling

• README

2. Contextual Embeddings

• README

3. Deep Reinforcement Learning

• README

3.1. Policy Gradient

3.2. Deep Q-Learning (DQN)

3.3. Actor-Critic

World\Stage	1	2	3	4
1
2
3
4
5
6
7
8

4. Conditional Generative Adversarial Nets

• README

4.1. Training Progress

• Resnet-based ACGAN with BCE loss (resnet_1000)

4.2. Human Evaluation Results

• Resnet-based ACGAN with BCE loss (resnet_1000)