Soroush Bateni


ABOUT ME

I am currently a third-year PhD student in the Department of Computer Science at the University of Texas at Dallas. I started my PhD studies in Spring 2018. My advisor is Prof. Cong Liu. I recieved my Master's degree in Computer Science from UTD in 2018.

My main area of research is in reliablity of autonomous systems with a specialty in DNN-driven autonomous driving. I have also worked on improving predictability of General Purpose GPU computing through better scheduling, better system software, and better performance/energy modeling.


soroush AT utdallas.edu
Google Scholar


My most recent work concerns resource management on autonomous embedded systems. This includes modeling the memory characteristics of unified memory architecture (RTAS 2020) and designing system software to cope with the disk pressure (RTAS 2020). Another set of work focuses on GPU-dependent Deep Neural Networks and their deployment on resource-restricted embedded hardware (USENIX ATC, RTSS 2018, RTSS 2018). Prior to that, I helped publish numerous papers at venues such as RTSS, RTAS, and TDPS.


News

  • Our paper titled "NeuOS: A Latency-Predictable Multi-Dimensional Optimization Framework for DNN-driven Autonomous Systems" has been accepted to USENIX Annual Technical Conference (ATC) 2020 (18.6% acceptance rate).
  • Our paper titled "Co-Optimizing Performance and Memory Footprint Via Integrated CPU/GPU Memory Management, an Implementation on Autonomous Driving Platform" has been accepted to RTAS 2020 (~27% acceptance rate).
  • Our paper titled "Predictable Data-driven Resource Management: an Implementation using Autoware on Autonomous Platforms" has been accepted to RTSS 2019 (~20% acceptance rate).
  • Finished summer Research Internship at Fujitsu Laboratories of America.
  • Two first-author papers have been accepted to the 39th RTSS (RTSS 2018): ApNet (exploring approximation potential of DNN for timing predictability), and PredJoule (a timing-predictable energy optimization framework for DNN-driven autonomous systems) (22.3% acceptance rate).
  • Our paper "S^3DNN: Supervised Streaming and Scheduling for GPU-accelerated Real-Time DNN Workloads" received the Best Paper Award at RTAS 2018 (CPSWeek@Porto).
  • Was the sole instructor for the Operating Systems Concepts (CS 4348) class in Summer 2018.
  • Our paper on "Exploring Computation and Data Redundancy via Partial GPU Computing Result Reuse" has been accepted to the 32nd ACM International Conference on Supercomputing (ICS 2018).
  • Recieved NSF ACM/IEEE RTSS Travel Grant 2018