Game AI

• Worked on an explainable machine learning method on credit report dataset. • Collaborated with computer science researchers to apply our proposed explainable AI method to an ML model.

• Proposed a novel interpretable AI method which makes neural networks more understandable to human users.• Analysis of the inner workings of the neural networks using Keras and TensorFlow frameworks.• Application of our explainable machine learning method on image classification tasks.• Worked on state-of-the-art convolutional neural networks such as ResNet, AlexNet, and VGG.• Focused on different image datasets such as ImageNet and Cifar10.• Designed a human subject study to evaluate our proposed method and analyzed the results using Rstudio.

• Worked as a team member on a project for RWDI consulting firm.• Designed and developed highly accurate ML models to predict a specificvariable in wind tunnels.• Dealt with geometric data extracted from over 120 constructions.• Used pyCaret library to develop and evaluate different ML algorithms such as xgboost, ridge and lasso regression.• Implemented python scripts to extract features from 3D building models in Rhinoceros software.• Applied data preprocessing and used different dimensionality reduction methods such as PCA and autoencoders.• Used different methods such as k-fold cross validation and grid search to perform hyper parameter tuning.• Implemented accurate deep neural networks using Keras and TensorFlow.• Created charts and plots in jupyter notebook to perform statistical analysis and visualized data using Matplotlib.