Carbon Nanotube Optoelectronic Synapse Transistor Arrays with Ultra‐Low Power Consumption for Stretchable Neuromorphic Vision Systems Tanghao Xie, Qinan Wang, Min Li, Yuxiao Fang, Gang Li, Shuangshuang Shao, Wenbo Yu, Suyun Wang, Weibing Gu, Chun Zhao, Minghua Tang, Jianwen Zhao

• This project focuses on optimizing mineral recovery circuits using a Genetic Algorithm approach, specifically for extracting the valuable mineral “gerardium.” By designing and evaluating various circuit configurations of separation units, the goal is to maximize recovery and purity of the final product while balancing economic considerations. The code was written in C++ and parallelized using OpenMP, and has been tested for large circuits on the HPC at Imperial.
• Achieved excellence in the final performance score.

• The project aims to improve emergency response protocols for hurricanes by forecasting the evolution of tropical cyclones in real-time using ML/DL techniques. Participants are tasked with designing and implementing ML/DL models to predict future satellite images and wind speeds based on provided data from 30 historical storms. The final solution has two models built with PyTorch: 1. Image Generation Model (Encoder+ConvLSTM+Decoder): generate future image predictions based on these existing images. 2. Prediction Model (Refined CNN-LSTM): predict wind speed based on the provided time series data.
• Achieved the best score in final prediction.

• Modelling the dynamics of airbursts of meteorites is a huge computational challenge. In this project we explored the effects and hazards caused by small asteroids as they enter the Earth’s atmosphere. Our team has developed an optimisation method for finding the asteroid radius and strength that gives the precise fit to an observed energy deposition curve for the Chelyabinsk meteor; developed a airburst solver capable of predicting the energy and location of explosions; developed an airburst damage mapper to plot damage zone and determine high-risk postcodes with user-friendly interface.
• Achieved the best performance in solver execution efficiency.

• INT303 Big Data Analysis (2223 S1) Assignment 2: Kaggle Competition - Will your employees leave?
• Final Score on Leaderboard: 12th (XJTLU_1929603)

• Based on a survey of students and teachers, a smart desktop ornament with integrated multiple functions was designed in order to optimize the online learning environment including simulation (TinkerCAD, Wokwi), physical prototype verification (based on Arduino, STM32 Development Board), and 3D modeling (Solidworks). Examiners can use wearable remote controller (ear-mounted controller with Bluetooth module) for remote proctoring, network connection module is based on python socket and an embedded AI voice recognition module for user interaction (voice commands).

• Achieve an outstanding score of 78%. Project Demonstration Video Available: link

• Python socket programming project: developed a reliable and efficient automatic file synchronization software based on P2P architecture that can be deployed on remote servers. Designed and built the protocol from scratch and implement larger file transfer (Test passed 2G large file transferring), file integrity monitoring (FIM), breakpoint transfer, and realized asynchronous transmission that enables near-complete utilization of LAN bandwidth.

• Achieved an excellent score of top 3% (final score of 93%) in the program stability and efficiency test (stress test) script written by module leader.