FastPath 2018 - 2018 International Workshop on Performance Analysis of Machine Learning Systems

The goal of FastPath 2018 is to bring together researchers and practitioners involved in cross-stack hardware/software performance analysis, modeling, and evaluation of machine learning systems. With microprocessor clock speeds being held constant, optimizing systems around specific workloads is an increasingly attractive means to improve performance. This approach is especially pertinent to the domain of machine learning, given the significant potential and widespread adoption of machine learning techniques recently. Machine learning systems have hardware and/or software specifically designed to run well for applications in the machine learning and AI domains. The types and components of such systems vary, but a partial list includes traditional CPUs assisted with accelerators (ASICs, FPGAs, GPUs), memory accelerators, I/O accelerators, hybrid systems, converged infrastructure, and IT appliances. The importance of machine learning systems is seen in their deployment in an ever-growing list of diverse systems including cellphones, high performance computing systems, database systems, self-driving cars, robotics, and in-home appliances. Exploiting hardware speed-ups for application level performance improvement poses several cross stack hardware and software challenges. These include developing alternate programming models to exploit massive parallelism offered by accelerators, designing low-latency, high-throughput H/W-S/W interfaces, developing techniques to efficiently map processing logic on hardware, and cross system stack performance optimization and hyperparameter tuning. Emerging infrastructure supporting big data analytics, cognitive computing, large-scale machine learning, mobile computing, and internet-of-things, exemplify system designs optimized for machine learning at large.
Topics
FastPath seeks to facilitate the exchange of ideas on performance analysis and evaluation of machine learning/AI systems and seeks papers on a wide range of topics including, but not limited to:
Workload characterization and profiling
GPUs, FPGAs, ASIC accelerators
Memory, I/O, storage, network accelerators
Hardware/software co-design
Measurements and experimentation
Performance modeling and prediction
Performance tooling and optimization
Techniques to accelerate data labeling
Performance of machine learning algorithms
Approximation techniques for machine learning
Power/Energy and learning acceleration
Programming models for machine learning systems
Runtime management systems
Workload scheduling and orchestration
Machine learning in cloud systems
Large-scale machine learning systems
Intelligent/cognitive system
Converged/integrated infrastructure
Machine learning systems for specific domains, e.g., financial, biological, education, commerce, healthcare