S P L A S H

FastAI Deep Learning - syllabus

Fast.AI Training in Chennai


FastAI is a high-level framework similar to Keras for Tensorflow. Fast.AI Implemented on top of PyTorch. FastAI "Top Down" learning approach Simplifying Coders to understand the Deep Learning framework. Easy to use and few lines of Deep learning code can build powerfull Computer Vison and NLP realtime application.

Fast.AI Overview

  • 1 Computer Vision/IMAGE - vision module of the fastai library contains all the necessary functions to define a Dataset and train a model for computer vision tasks.
  • 2NLP/TEXT - The text module of the fastai library contains all the necessary functions to define a Dataset suitable for the various NLP (Natural Language Processing) tasks and quickly generate models you can use for them.
  • 3Tabular - Predictive analytics, Classification on numerical data. tabular contains all the necessary classes to deal with tabular data.
  • 4ULMFIT - Universal Language Model Fine-tuning for Text Classification, is one of the fisrt transfer learning language model for NLP specific tasks.
  • 5 Deployment/ - Multiple ways to deploy your trained models Render, Google Cloud, AWS, MS Azure and Dockers Kubernetes
  • 6 Data block API - Datasets can be messy. The data block API is designed to be extremely flexible to handle data from different source. The data block API helps coders turn our data items into a DataBunch that is read to use in fastai models.

Fast.AI DeepLearning - Training Syllabus

 

Module 1 – Solving Computer Vision problem

  • Image Classification - Code walkthrough
  • How to prepare Image data
  • live project : Image classification with any realworld datasets
  • Creating your own image dataset
  • Learning PyTorch
   

Module 2 – Prediction and Classifioan on Tabular data

  • Build your classification model with categorical data
  • Fine tuning parameters and Improve the accuracy
  • Hands-on : Bring your data & build classification model
  • Predicve analytics with continuous data
  • Live project : Build and deploy model

Module 3 - Natural Language Processing

  • Understand Transfer Learning
  • ULMFiT model for Text classification
  • Hands-On : Next word Predictor and Sentiment analysis
  • Live Project : Classify Emotion/Sentiment in reviews

Module 4 – Advanced training techniques

  • Hand-on : Stochastic Gradient Descent (SGD) optimizer
  • Live project : Back propagation
  • Loss functions, optimizers, and the training loop
  • Hands-on : importance of learning rate(LR)
  • Problem of overfitting

Module 5 – Building Recommendation system

  • Collaborative filtering
  • Data Cleansing
  • Hyperparameter tuning- Regularization, Loss function.
  • Live Project : Building Movie Recommendation using IMDB Dataset

Module 6 - Fast.AI Deep Learning Deployment & Others

  • How to - Deploy model with Cloud platforms
  • Deep Dive into Neural Net
  • Adam Optimizer - Overview
  • Production ready Application

Fast.AI and PyTorch upstart framework gaininig momentum with its rich ecosystem tools. FastAI library simplifies training fast and accurate neural nets using modern best practices. Easy to build and deploy Deep Learning model Making neural nets uncool again.

PyTorch alternatives Alternative of with TensorFlow still dominating world market Interested in TensorFlow?

SPLASH - Aim to focus DATA - Training, consulting and Projects implementation. check out our list our data training courses.