LDAWordCloud | Topic Visualization and Application
Developed a PYTHON library to visualize and evaluate topic models generated from LDA(Latent Dirichlet Allocation) method.
• Generated word clouds for LDA models by word distribution for topics and number of topics assigned for the top words in the distribution.
• Applied topic model evaluation to nature language processing tasks including email classification
DodgeBird | Motion Sensing Game on FPGA
Implemented video streaming, UI rendering and game logic on a FPGA device with Vivado and Vivado HLS.
• Realized MIMO and AXI protocol between CPU and FPGA.
• Simulated real time register-memory data transmission under different data paths
• Improved a gesture recognition algorithm for user input to minimize the error within 15 pixels and to accelerate the game from 4FPS to 26FPS with hardware accelerating methods including parallel computing, pipelining and array partition.