Workshop
Workshop: Building Deep Learning Solutions from Scratch with Keras
Deep Learning models are dominating nowadays in a variety of application domains and have outperformed the classical machine learning models in many ways. It is always a curiosity across the developers and learners on how to build efficient deep learning models for interesting applications. In this workshop, developing deep learning models from scratch will be discussed through which the participants will learn how to start, develop and apply these models in real-life applications. These implementations will be done with Keras using the TensorFlow backend.
30th Oct at Deep Learning DevCon 2020
One day virtual workshop from 10am to 5pm
Curriculum
-
- Introduction to Deep Learning
-
-
- Introduction
- Machine Learning Vs Deep Learning
- Applications of Deep Learning
- Introduction to deep learning frameworks.
-
- Feed Forward Neural Networks
-
-
- Introduction to Neural Networks
- Learning Algorithms for Neural Networks
- Supervised Learning with Neural Networks
- Backpropagation
- Optimization and Gradient Descent
-
- Deep Neural Networks
-
-
- Simple Deep Neural Network
- Deep Neural Network in Classification
- Convolutional Neural Networks (CNN)
- CNN for Image Classification
- Variants of Convolutional Neural Networks
- Advanced applications of CNN in computer vision
-
- Improving Deep Learning Models
-
- Regularization Techniques
- Optimization Algorithms
- Batch Normalization
- Hyperparameter Tuning
- Recurrent Neural Networks
-
-
- Introduction to Recurrent Neural Networks (RNNs)
- Different types of RNNs
- Long Short Term Memory (LSTM) RNN
- Vanishing Gradient Problem
-
- Sequence Modeling
-
-
- Word Representation
- Word Embedding
- Word2Vec
- Text Classification
- Text Generation
-
- Autoencoders
-
-
- Introduction to Autoencoders
- Deep Autoencoders
- Convolutional Autoencoders
- Image Reconstruction
-
- Generative Adversarial Network
-
-
- Introduction to GANs
- Generator and Discriminator
- Fake Image Generation
-
- Neural Style Transfer
-
-
- Concept of Neural Style Transfer
- Style transfer using Convolutional Neural Network
- Style Transfer using TensorFlow HUB (TF-HUB) Module
-
- Real-Time Object Detection
-
- How to detect objects?
- Image segmentation and Instance segmentation
- Frameworks for object detection
- Object Detection in images
- Object detection in videos
Prerequisites
- Understanding of Vectors, Matrices and Tensors.
- Knowledge of Python programming language and the basic idea of TensorFlow and GPU.
- Basic understanding of machine learning algorithms.
- Good internet connection to load the datasets during program execution.
- Familiarity with Google Colab
Required Tools
- Google Colab
- If working in Jupyter notebook or Spyder editor of Anaconda:-
- NumPy, Pandas, TensorFlow, Keras libraries pre-installed
- High-speed Internet connectivity
Learning Outcomes
Strong understanding of deep learning, from basics to advanced.
In-depth knowledge of artificial neural networks.
Clear concept about development, tuning, regularizing and improving the models.
Knowledge of various building blocks with their practical implementations.
Understanding of real-life applications of deep learning models.
Practical knowledge of applying deep learning in computer vision and natural language processing.
Passes
Register before the passes get over.- Access to all tracks & workshops
- Access the recorded sessions later
- Free
Don’t forget:
OCT
