AutoML tutorial

MICCAI 2022 Tutorial on

Automated Machine Learning and Neural Architecture Search

Overview

Deep neural networks (DNNs) have been successfully applied to volume segmentation and other medical imaging tasks. They are capable of achieving state-of-the-art accuracy and can augment the medical imaging workflow with AI-powered insights.

However, training robust AI models for medical image analysis is time-consuming and tedious and requires iterative experimentation with parameter tuning. On the other hand, automated machine learning (AutoML) has been studied and developed for years in academia and industry. Its objective is to construct AI models without the need for human heuristics.

This tutorial of the most recent advancement of AutoML aims to enable researchers and data scientists with cutting-edge tools for AI development in medical image analysis.

The AutoML functionality makes the process of neural architecture search and hyper-parameter tuning seamless by intelligently searching for the optimal parameter settings to train models automatically. The underlying algorithms give data scientists a configurable environment to define the training workflow experiments. Moreover, it provides a standard way to implement state-of-the-art deep learning (DL) solutions.

Learning Objectives

Understanding the benefits of applying AutoML and NAS techniques for medical imaging research, and the importance of AutoML and NAS in modern research development.

Learning the current advances of AutoML and NAS in medical imaging and other related research fields (e.g., machine learning, computer vision, etc.).

Applying the existing technologies of AutoML and NAS to existing research projects for performance improvement.

Analyzing potential research scenarios to determine whether to apply AutoML/NAS algorithms.

Perceiving the opportunities to improve current AutoML and NAS algorithms and launch related research projects under medical imaging contexts.

Talks

Topic			Speaker
Opening	TBD	TBD	TBD
Talks
AutoML for Data-Efficient Medical Image Segmentation	TBD	TBD	Dongnan Liu (The University of Sydney)
NAS in Medical Imaging Segmentation with its Applications to Head & Neck Organ at Risk and Thoracic Lymph Node Station	TBD	TBD	Dazhou Guo (Alibaba)
Model and Data Efficient Deep Learning for Computer Vision	TBD	TBD	Bichen Wu (Meta)
GPUNet: Searching the Deployable Convolution Neural Networks for NVIDIA GPUs	TBD	TBD	Lingnan Wang (NVIDIA)
AutoML Best Practice with MONAI for Medical Imaging	TBD	TBD	Dong Yang (NVIDIA)
Closing	TBD	TBD	TBD

Useful Links

Medical Open Network for Artificial Intelligence (MONAI): https://monai.io/
Literature on Neural Architecture Search: https://www.automl.org/
Literature on Neural Architecture Search in Medical Domain: https://www.automl.org/
NVIDIA Tensor Cores: https://www.nvidia.com/en-us/data-center/tensor-cores/
Automatic Mixed Precision (AMP): https://nvidia.github.io/apex/amp.html
PyTorch AMP: https://pytorch.org/docs/stable/amp.html
TensorFlow AMP: https://developer.nvidia.com/blog/nvidia-automatic-mixed-precision-tensorflow/
High-Performance Sample Code for Various Applications: https://github.com/NVIDIA/DeepLearningExamples

Organizers


Dongnan Liu	Daguang Xu	Dong Yang	Yefeng Zheng	Zhuotun Zhu