ramayzoraiz Goto Github PK

p001-single-neuron-scratch

Binary classification by single vanilla neuron (logistic regression inspired). Neuron is implemented using python without frameworks.

p002-single-neuron-scratch-optimization

Optimization of single vanilla neuron(logistic regression inspired); and run times are provided for binary classification task. Numpy vectorization and numba are used to speed up code.

p003-neuron-scratch-pytorch-tensorflow

p004-neural-network-1-hidden-layer

Classification is done by 1 hidden layer neural network; implemented in vanilla python, tensorflow and pytorch. In tensorflow/pytorch, different APIs are explored to implement same algorithm.

p005-neural-network-l-layers

Binary classification is done by "L" layer neural network; implemented in vanilla python, tensorflow and pytorch. In tensorflow/pytorch, different APIs are explored to implement same algorithm.

p006-neural-network-layers-comparison

Using synthetic data for binary classification, explore neural networks layers effect and limitation. Implemented in tensorflow and pytorch.

p007-multiclass-classification-l-layered-neural-network

Using synthetic data, multiclass classification is done by "L" layers neural network; implemented in tensorflow and pytorch.

p008-lenet-cnn-model

LeNet model and its variations is trained on OxfordiiiT pet dataset. Effect of L2 regularization and BatchNorm in tuning model is observed.

portfolio_projects

kkk