This repository contains materials (notebooks and a presentation) associated with "AI Workshop with Semiotic Labs" organized @ Edge&Node's mindblowing House of Web3 in Presidio of San Francisco on May 5th 2023.
The workshop will consist of two parts. In the first part we will do a brief introduction, explaining what we understand as intelligence and what artificial intelligence is. We will do a brief history of how AI emerged, we will explain what the domains of AI are and what Machine Learning, Reinforcement Learning and Deep Learning are. We will focus then on the latter, go through some of the most important turning points in the history of Neural Networks, from Rosenblatt’s perceptron up to OpenAI’s ChatGPT that recently became a phenomenon on a global scale. We will summarize the talk by high-lighting selected emerging problems and future research directions.
In the second part we will do a hands-on tutorial, focusing on two hello-world deep-learning tasks. In the first task we will focus on the Computer Vision (CV) domain and build and train a simple neural network able to recognize images. In the second part we will implement one of the Natural Language Processing (NLP) classical tasks and build a text classifier able to distinguish different types of news.
An elementary knowledge of Python will be required. Understanding of Pandas and PyTorch packages is not required, but for sure will be useful. Both hands-on tutorials are based on Google Colab.
This workshop addresses the audience interested, but not very familiar with Machine Learning/Deep Learning. Along with the world-wide adoption of Generative AI techniques and applications, most people focus now on the prompting techniques, treating the underlying neural network models as black boxes. The goal of this workshop is to dive a little bit deeper and enable participants to develop intuition through hands-on experience with the models.
- 11:00am-11:05pm - Intro: About Semiotic Labs
- 11:05am-12:00pm - Talk: “Evolution of multi-modal AI”
- 12:00pm-12:45pm - Lunch break
- 12:45pm-1:30pm - Tutorial I: Image Classification on CIFAR-10 (CV)
- 1:30pm-1:45pm - Coffee break
- 1:45pm-2:30pm - Tutorial II: Text Classification on AG_NEWS (NLP)
- 2:30pm-2:45pm - Closing remarks
(sorry, the slides are barely visible on the recording, but hey! here are the slides)
All the code/notebooks are distributed under Apache v2 license.
The presentation is distributed under CC BY 4.0 license.
All the opinions expressed by the author (Tomasz Kornuta) are of his own and do not represent Semiotic Labs, Edge&Node, House of Web3 nor The Graph Protocol.