This project explored using various learning algorithms to predict trial outcomes using the transcripts of the Old Bailey proceedings. The dataset comprised of 25,000 trials labeled as guilty or not guilty. Five classifiers were implemented from scratch, including ID3, Perceptron, SVM with SSGD, logistic regression with SGD, and bagged ID3, and a neural network was implemented using TensorFlow/Keras. Different feature representations like bag-of-words, tfidf, glove embeddings, and categorical attributes were examined. Neural networks achieved the best performance (0.83 on Kaggle), highlighting their suitability for nonlinear data. Other algorithms performed comparably (0.79-0.80 on Kaggle). The project emphasized the importance of algorithm selection, feature engineering, representation, and hyperparameter tuning in machine learning.
Analyzed two methods for quantifying the predictive uncertiainty of DeformerNet, a framework that provides a learning-based approach for minipulating deformable objects to a desired goal shape
Visualized air quality data collected from research-grade pollutant sensors on light-rail cars and e-buses (courtesy of the Department of Atmospheric Sciences at the University of Utah) in conjunction with socioeconomic data by ZCTA to underscore the disparities in air quality exposure and the environmental injustice for those living in the west side of the Salt Lake Valley
Developed finite element models that represent the tibiotalar joint with different articular morphologies to investigate the biphasic-on-biphasic contact in healthy and progressively osteoarthritic conditions
Implemented Monte Carlo simulation in a game show given choice of three doors, a car behind one and goats behind others
Examined NMPC stability using Lyapunov Theory; tuned PID controller for leg extension in musculoskeletal model
Utilized MATLAB Ode45 to model isometric muscle force; performed sensitivity analysis on physiological parameters in force and fatigue prediction
Built CNN to identify if a remotely sensed object was a ship or iceberg in Kaggle competition
Designed and built a mobile robot with 5 DOF manipulator -- 3D printed parts, DC motors, LIDAR, IMUs, Raspberry Pi, and microcontrollers; derived inverse kinematics for block picking; STOMP algorithm
Executed iLQR algorithm for object reaching task in biomechanical human arm model moving in sagittal plane
Expanded stochastic trajectory optimization with LQR feedback control algorithm for mobile robot motion planning