Projects related to machine learning, deep learning, etc.
-
House price prediction - Model to predict the price of house given its year of sale, age, location, etc. Implemented dense neural netowrk with three hidden layers and output layer. Used tensorflow (keras) framework with adadelta optimizer and earlystopping callback.
-
Basic image classification - Model to classify the input images of handwritten numbers (MNIST) from 0 to 9. Implemented dense neural network with two hidden layers and output layer. Used tensorflow (keras) framework with sochastic gradient descent and one hot encoding.
-
Linear Regression with Numpy - Model to apply linear regression to map n input features to output using numpy. Used gradient descent and implemented propagation and optimization from scratch.
-
Image classification CNNs with Keras - Model to apply convolutional neural networks to classify images into three classes (can be extended to any number of classes) using keras. Used adam optimizer and created convolution, maxpooling and fully connected layers. Also applied batch normalisation and dropout regularization.
-
Traffic sign classification - Model to apply convolutional neural networks to classify 43 types of traffic signs using keras. Used adam optimizer and created convolution, averagepooling and fully connected layers. Also applied dropout regularization.
-
Handwritten numbers recognition - Model to apply convolutional neural network with deep neural network to classify handwritten numbers (MNIST). Used Keras API with adam optimizer.