Aditya Dewan's repositories
Adam-Optimization-From-Scratch
📈Implementing the ADAM optimizer from the ground up with PyTorch and comparing its performance on six 3-D objective functions (each progressively more difficult to optimize) against SGD, AdaGrad, and RMSProp.
Born-Again-NNs-From-Scratch
🤝 Implementing BANs with Teacher Confidence Weighting and Dark Knowledge Permutations from scratch, and testing said implementations between DenseNet-121 teachers and ResNet-18 student ensembles on the ImageWang Dataset (along with some MLP MNIST tests).
Crysta-MVP
The code for a company I built - using data to improve student energy levels 👀
pytorch-hessian-eigenthings
Efficient PyTorch Hessian eigendecomposition tools!
SADDLE-POINT-RECRUITMENT-FOR-KNOWLEDGE-DISTILLATION
ISEF 2023 (TEAM CANADA) PROJECT. Find the complete documentation and code in the README file linked here and below.
Symptom-Diagnosis-AI
A project I'm currently working on that intakes different symptoms and provides a diagnosis. The model currently has 92% accuracy - check out a demo in the webapp folder!
Custom-Maxout-Activation-Implementation
💡Implementing a custom Maxout network from scratch (as an extension of nn.Module in Pytorch) and testing model performance on MNIST in comparison to ReLU networks to determine whether Maxout's more complex function approximations can provide higher accuracies in real-world use cases.
Autonomous-Vehicle-Simulator-Behavioural-Cloning
Simulating Autonomous Vehicles with Behavioral Cloning and CNNs!
HOUSING-PRICE-PREDICTION-XGBOOST
FINAL ECONOMICS SUMMATIVE - EMPLOYING GRADIENT-BOOSTING DECISION TREES FOR THE PREDICTION OF HOUSING PRICES: A HEDONIC-BASED ANALYSIS OF CENTRAL FACTORS IMPACTING RESIDENTIAL HOUSING.
knowledgeDistillation
PyTorch implementation of (Hinton) Knowledge Distillation and a base class for simple implementation of other distillation methods.
MarkovChainDemo
Generate text using a Markov chain
PyHessian
PyHessian is a Pytorch library for second-order based analysis and training of Neural Networks