deep learning
Deep Learning Basics
dive into the fundamentals of neural networks including Convolutional Neural Networks, Artificial neural networks, and recurrent neural networks
Fundamental Packages
get hands-on experience with the two most popular Python libraries: Tensorflow and Pytorch, and understand when to use each
skills + applications
take deep learning beyond the intuition and tools by applying these models to real-world scenarios, learning best practices for training, re-training, and sustainable deployment
OVERVIEW
This workshop will introduce the fundamentals of deep learning from the ground up.
Find out what Deep Learning is, why it is different (and better!), and how to use it effectively.
The workshop will cover several interesting applications along the way, including computer vision and natural language processing.
Practical notebooks using the PyTorch framework will also be covered, demonstrating how you can easily apply it to your problems.
EXPERIENCE LEVEL
AUDIENCE
STEPHEN WELCH
VP of Data Science, Mariner
Stephen Welch is VP of Data Science at Mariner, where he leads a team developing deep-learning based solutions for manufacturing applications. Prior to working with Mariner, Stephen was VP of Machine Learning at Autonomous Fusion, an Atlanta-based autonomous driving startup, where Stephen lead the design, development, and deployment of machine learning algorithms for autonomous driving.
Stephen has extensive experience training and deploying machine learning models across a wide variety of domains, including an on-board crash detection algorithm that is now deployed in over 1M vehicles as part of the Verizon Hum product. Stephen strives to not just develop strong technology, but to explain and communicate results in clear and accessible ways – as an adjunct professor at UNCC, Stephen teaches a 60+ person graduate level class in machine learning and computer vision.
Stephen is the author of the educational YouTube channel Welch Labs, which has earned 240k+ subscribers and 10M+ views. Stephen holds 10+ US patents, and engineering degrees from Georgia Tech and UC Berkeley.
AGENDA
7:30 AM - 4:00 PM
REGISTRATION + BREAKFAST
MORNING SESSION
Coffee Break Included in Each Session
| Sprint | Key Topics |
| A Brief History of Neural Networks | Perceptrons, multilayer Perceptrons, neural networks, the rise of deep learning |
| Let's Get to Know Each Other | Quick audience poll + chat, what are you looking to get out of this workshop? |
| Brief Overview of How We'll Be Using Jupyter, Python, Pytorch, and JupyterLab, and fastai | Getting your environment setup, install + test packages. |
| Coffee/Stretch Break | |
| Neural Networks Demystified | The mechanics and mathematics for forward and backpropagation in neural networks. Overfitting + Regularization. |
| [Bonus Session if Time] State of the Art Deep Learning for Computer Vision | State of the art of deep learning for computer vision. |
LUNCH + NETWORKING
AFTERNOON SESSION
Coffee Break Included in Each Session
| Sprint | Key Topics |
| A brief introduction to Pytorch | Pytorch as "Numpy with GPU Support”, simple neural network in Pytorch, automatic differentiation, nn.Module, PyTorch layers, PyTorch Optim, nn.Sequential |
| How to Build a World Class Deep Learning Model [Part 1] | Stochastic gradient descent, regression vs classification, one hot encoding, cost functions and maximum likelihood, cross entropy |
| Coffee/Stretch Break | |
| How to Build a World Class Deep Learning Model [Part 2] | CNNs, pooling and strides, AlexNet walkthrough, ImageNet, transfer learning, adaptive pooling, dropout, data augmentation, a little historical perspective |
| Get results fast with fastai | Jeremy Howard and the fastai philosophy, DataBunches, Learners, NLP with fastai, world class computer vision with fastai |
| [Bonus Session if Time] GANs | Ian Goodfellow invents GANs, the world's simplest GAN & nash equilibria, a dive into higher dimensions, DCGAN to the rescue, Visualizing GANs, GAN grow up (sortof), StyleGAN insanity, the unbelievably interesting world of GAN variants |