Workshop on Reproducible and Pareto-Efficient Deep Learning
1st Workshop/Tournament on SW/HW Co-Design of Pareto-Efficient Deep Learning (ReQuEST)
co-located with ASPLOS’18
Williamsburg, VA, USA
March 24th, 2018 (afternoon)
1) Keynote by Prof. Yiran Chen (Duke University, USA):
“The Retrospect and Prospect of Low-Power Image Recognition Challenge (LPIRC)”
2) 5 presentations of image classification algorithms with reproduced workflows and shared plug&play artifacts covering diverse technology (TensorFlow, Keras, Avro, Caffe, MXNet, NNVM, TVM, ArmCL, OpenBLAS, CUDA, OpenCL), 32/16/8 bit precision models (VGG16, AlexNet, ResNet-50, ResNet-18, Inception-V3, SSD, autotuned MobileNets), and platforms (AWS, Raspberry Pi, Arm, NVIDIA Jetson TX2, Intel Xeon servers, Firefly-RK3399, HiKey 960 GPU)
3) Announcing final ReQuEST scoreboard accompanied by a repository with above portable, customizable and reusable AI/ML components
3) Open discussion panel: “Tackling complexity, reproducibility and tech transfer challenges in a rapidly evolving AI/ML/systems research”
Preliminary list of participants:
– Yiran Chen, Duke University
– Charles Qi, Cadence
– Tianqi Chen, University of Washington
ReQuEST tournaments and workshops bring together multidisciplinary researchers (AI, ML, systems) to find the most efficient and reproducible solutions for realistic problems requested by the advisory board in terms of speed, accuracy, energy, complexity, costs and other metrics across the whole application/software/hardware stack.
All the winning solutions (code, data, workflow) on a Pareto-frontier are then available to the community as portable and customizable “plug&play” AI/ML components with a common API and meta information.
The ultimate goal is to accelerate AI/ML research and reduce costs by reusing the most accurate and efficient AI/ML blocks continuously optimized and autotuned across diverse models, data sets and platform from the cloud to edge.