Duration
4 days
Pre-requisites
Knowledge of Machine Learning and Python is required.
Course Objectives
- To make participants capable of programming common machine learning algorithms in python.
- To make participants understand possible use cases for AI/ML/DL.
- To make participants capable of programming common Deep learning algorithms in python.
- The students will be able to take up any kind of input data, structured or unstructured and process it for machine and deep learning.
- To make participants understand the practices of working with data to enable creation of AI systems from data.
Course Outcome
By the end of the course the participants will be able to derive insights for building an AI system from any kind of data using machine learning and deep learning algorithms.
Course Outline
Basics of Deep Learning
- Introduction to Deep Learning and Neural Networks
- First look at Neural networks
- Backpropagation
- Activation Functions
- XOR – Inability
- Multi Layer Perceptron
- SKLearn
- Concept of Gradient Descent
- Applications of Neural Networks and Deep Learning Intro to Tensorflow
- Lazy evaluation
- Programming neural network with Keras + Tensorflow (Simple Case Study)
Computer Vision
- Computer vision
- What is Digital Image Processing and Computer Vision
- Applications of Computer Vision
- Basic Image Transformations
- Image Feature extraction
- HOG, and Other feature extraction
- Extraction of Image from Video
- Viola Jones Face detection
Convolutional Neural Networks
- Introduction to CNN
- Neural Network (CNN)
- Image classification with CNN
- Deeper dive into concepts of convolutions, strides and paddings, pooling layers, dropouts
- Case study of Image Classification in Keras with CNN (Keras with Tensorflow)
Recurrent Neural Networks and Sequence Data Processing
- Introduction to RNN
- Backpropagation in RNNs
- Problems with vanishing Gradients
- Basic RNN
- LSTM architecture
- GRU architecture
- Applications of RNN
- Case study of demand forecasting with RNN
Natural Language Processing
Text Mining Concepts:
- WordCloud,
- Term Document Matrix,
- CoOccurrence& Concordance
- Named Entity Extraction
- POS tagging
- Word Embedding’s
- Word2Vec and Glove
- Text Classification for Sentiment with Word Embedding’s
- Recent Work on Attention Models for Text
Programme scope, delivery arrangements and suitability are confirmed during enquiry.