Machine learning (ML) and artificial intelligence (AI) algorithms, in general, are not readily available in package format, with a few exceptions, e.g., NLTK Vader and TextBlob Sentiment. Unfortunately, these available packages often turn out to be quite underwhelming in terms of their accuracy and their utilities are limited to highly specific tasks. For customized machine learning and deep learning needs, one must create and train one’s own models to obtain explainable and reliable results.

Creating and training one’s own AI model from scratch is a daunting task, with potential pitfalls of incorrect algorithm implementations and slow models. An important insight on AI models’ computational graphs, is that AI models contain many repeated and reusable modules, e.g. regularizations, activation functions and loss metrics. This insight leads to the development and popularity of AI packages, TensorFlow, Torch, Trax, etc. These AI packages provide highly optimized modules, which makes creating an efficient AI neural network model as simple as a few lines of code. This workshop will introduce you to one of the most popular AI packages, TensorFlow, in Python. By the end of this workshop, we will build and train a few AI models together!

You should walk away from this workshop with the ability of creating your own AI models!

Objectives:

Participants will learn:

How to describe machine learning and deep learning models in terms of numerical models
How to represent AI models using computational graphs
What are computational graphs and how to optimize computational graphs
Building blocks of AI models
Hands-on experience of building your own image recognition AI model
Hands-on experience of building your own natural language processing model
Hands-on experience of building your own time series forecasting models

Level: Intermediate

Duration: 2 hours

Prerequisites: TensorFlow package contains many program interfaces (Python, C++, Java, etc). This workshop will use Python API for simplicity and for tutorial readability. It would be great if attendees are familiar with basic Python programming, but it’s not required. The focus of this workshop is on how to build AI models.

Github repository: Will be made available later

Introduction to Tensorflow