Exascale Deep Learning to Accelerate Cancer Research

TL;DR

This paper presents an HPC-enabled method for automatically generating neural network architectures optimized for both accuracy and speed, significantly accelerating cancer pathology data analysis for researchers with limited computational resources.

Contribution

It introduces MENNDL, a neural architecture search framework that produces tailored, faster neural networks without sacrificing accuracy, suitable for large-scale cancer data analysis.

Findings

Achieved comparable accuracy to state-of-the-art networks on cancer datasets.

Generated neural networks that are 16 times faster at inference.

Enabled researchers with modest resources to analyze data in real-time.

Abstract

Deep learning, through the use of neural networks, has demonstrated remarkable ability to automate many routine tasks when presented with sufficient data for training. The neural network architecture (e.g. number of layers, types of layers, connections between layers, etc.) plays a critical role in determining what, if anything, the neural network is able to learn from the training data. The trend for neural network architectures, especially those trained on ImageNet, has been to grow ever deeper and more complex. The result has been ever increasing accuracy on benchmark datasets with the cost of increased computational demands. In this paper we demonstrate that neural network architectures can be automatically generated, tailored for a specific application, with dual objectives: accuracy of prediction and speed of prediction. Using MENNDL--an HPC-enabled software stack for neural architecture search--we generate a neural network with comparable accuracy to state-of-the-art networks on a cancer pathology dataset that is also $16\times$ faster at inference. The speedup in inference is necessary because of the volume and velocity of cancer pathology data; specifically, the previous state-of-the-art networks are too slow for individual researchers without access to HPC systems to keep pace with the rate of data generation. Our new model enables researchers with modest computational resources to analyze newly generated data faster than it is collected.