SumitxThokar / Neural-Networks

Neural Networks Programs from Zero.

Neural Network

A neural network can be defined as a computational model inspired by the structure and function of the biological neural networks, consisting of interconnected nodes (neurons) that process and transmit information by performing simple computations. These computations are determined by the weights assigned to the connections between neurons and the activation functions applied to their outputs. Neural networks can learn patterns and make predictions based on input data, and are widely used in various applications such as image classification, speech recognition, and natural language processing.

Hello World of Neural Network

• The very first model program of Neural Network. The model used a single layer neural network built with TensorFlow to analyze the predictions.
• Jupyter Notebook file Here

Very First Computer Vision Program

• Got introduced to the field of Computer Vision. It's all about teaching a computer to understand and label what is present in an image. I worked on my first computer vision project using the Fashion MNIST dataset, which comprises of 70,000 images and 10 categories, all with a resolution of 28x28. I was thrilled to train a neural network with three layers and see the results for myself.
• Jupyter Notebook file Here

Callback

• Learned about callback and its function in controlling the training process. The callbacks API allows you to stop the training process when a specific metric is reached, saving time and resources.
• Jupyter Notebook file Here

Convolutional Neural Networks (CNNs)

• Introduced to CNNs and used them to build a much better fashion classifier. Also got the concepts of convolutional and pooling and successfully implemented them.
• Jupyter Notebook file Here

Exploring Convolutions

• Explored how convolutions work by creating a basic convolution on 2D grayscale image. Also effect of (2,2) Max Pooling.
• Jupyter Notebook file Here

Convolutional Neural Networks in Tensorflow

• Worked on Cats vs Dogs datasets.
• Built a model to classify real images of cats and dogs.
• This model is the first one we build we using previous ideas and this model is overfitted.

C2W2 Assignment Project on Popular Cat vs Dog dataset from kaggle

• Used the concept of Image Augmentation.
• Project Here

C2W3 Assignment Project on Horse vs Human dataset.

• Used concept of Image Augmentation, Transfer Learnning and Dropout.
• Project Here

Natural Language Processing in Tensorflow.

Predicting the next word.

• Implemented model and train it using corpus of Shakespeare's sonnets, while also creating some helper functions to pre-process the data.
• Project Here