School of Engineering Department of Computer Science and Engineering 126 Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: TSUI Yuk Hang / MATH Course: UROP1100, Spring The Deep Convolutional Generative Adversarial Network (DCGAN) model has a huge contribution in generating high degree realism images with large latent dimensions, which is not favorable for end-user to control and generate the images they want without the understanding of the meaning of each dimension. We combine the conditional DCGAN, Variational Auto-Encoder (VAE), and Principal components analysis (PCA) to create a User-Interface (UI) with low dimensions for generating realistic images. The research study aims to provide a method for generating high dimensional images through a low dimensional input to benefit end-user control and study how the deep learning network interprets the image generation process. Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: ZHAO Yizhe / COSC Course: UROP2100, Spring Human-in-the-loop means that humans train, test, or tune the AI system to help it achieve more reliable results. Human-in-the-loop can be performed to teach the system what features to look for and guide the system to get more accurate answers. In this project, we learn the concept of Offline Reinforcement Learning and Class Activation Maps. We implement a behavior cloning agent on Atari Breakout game and train it via class, then use CAM to visualize the agent understanding. We are using interpretable ML technique to show the process of learning and try to provide the visualization during learning then game AI/Robot trainer can adjust the model according to the learning progress better instead of only relying on the loss function. Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: ZHOU Xinrui / COGBM Course: UROP2100, Spring This project aims at exploring explainable machine learning and different interpret methods. The main tasks for this semester are to study related XAI topic and interpret methods for machine learning models and their different variations, then used Captum package to do interpretation on CNN and BERT model for text classification based on word importance and corresponding sentiment. Also, we reviewed many some related papers that committed to this field. The research assistant Mr. Arthur Yau led weekly meetings where students presented what they have learned and implemented for the week. The main task for students were studying the materials provided by the research assistant and also implemented and reproduced some paper results and do some literature reviews.