C-BRIC Researchers Present at DAC Conference
C-BRIC is proud to have many researchers share their work at the Design Automation Conference (DAC). DAC is the leading event for “the design and design automation of electronic chips to systems. DAC offers outstanding training, education, exhibits and superb networking opportunities for designers, researchers, tool developers and vendors.” DAC was held July 10-14 in San Francisco, California, at the Moscone Center West conference center.
C-BRIC faculty Arijit Raychowdhury (Georgia Institute of Technology) and Kaushik Roy (Purdue University) presented as part of the Semiconductor Research Corporation (SRC) Joint University Microelectronics Program (JUMP) session on Microelectronics and Computing in 2030: A co-design perspective. Raychowdhury presented “A Systems driven approach to Semiconductor Research and Innovation” in his role with the JUMP ASCENT Center along with his ASCENT colleagues. Roy, C-BRIC Director, presented “A Cross-layer Approach to Cognitive Computing.”
Raychowdhury participated in the “Cryogenic Computing, Super Cool or Not?” panel discussion as well. Roy and C-BRIC Intel liaison Arnab Raha were part of the “Approximate Computing, Fiction or Reality?” panel. And C-BRIC faculty Priya Panda (Princeton University) co-led the AI and ML on next generation computing platforms session.
C-BRIC also had students honored as part of the DAC Young Fellow program. Those selected for the Young Fellow program included Soumendu Ghosh (Purdue University), Foroozan Karimzadeh (Georgia Institute of Technology), Abinand Nallathambi (Purdue University), Aradhana Parvathy (Purdue University), Deepika Sharma (Purdue University), and Zishen Wan (Georgia Institute of Technology). Ghosh presented “Approximate Inference Systems (AxIS)” and was a winner of the best 20 videos of the Young Fellow program. Karimzadeh presented “BitS-Net: Bit-Sparse Deep Neural Network for Energy-efficient RRAM Based Compute-In-Memory.” Nallathambi presented "Layerwise Disaggregated Spiking Neural Networks." Parvathy presented "Sequence-based Approximations for Compressing Ultra-low precision Deep Neural Networks." Sharma presented “Identifying Efficient Dataflows for Spiking Neural Networks.” And Wan presented “Analyzing and Improving Computing System Reliability for Safety-Critical Autonomous Drones.”
Additional works from C-BRIC researchers included:
Amrit Nagarajan, Jacob Stevens, and Anand Raghunathan from Purdue University presented “Efficient Ensembles of Graph Neural Networks” as part of the New Algorithm Design and System Optimization for Machine Learning Methods session.
Fan Zhang, Li Yang, Jian Meng, Jae-sun Seo, and Yu Cao of Arizona State University, along with collaborator Deliang Fan, presented "XMA: A Crossbar-aware Multi-task Adaption Framework via Shift-based Mask Learning Method" in the Do Not Forget the Software: Bare Metal Neural Acceleration is no fun without it session.
Sarada Krithivasan, Sanchari Sen, Nitin Rathi, Kaushik Roy, and Anand Raghunathan of Purdue University presented "Efficiency Attacks on Spiking Neural Networks" in the Preventing Brain Drain: How to Secure Next Generation AI session.
Shubham Negi, Indranil Chakraborty, Aayush Ankit, and Kaushik Roy of Purdue University presented “NAX: Neural Architecture and Memristive Xbar based Accelerator Co-design” as part of the Domain-Specific PIM Accelerators from Client to Cloud session.
Abhiroop Bhattacharjee, Yeshwanth Venkatesha, Abhishek Moitra, and Priyadarshini Panda of Princeton University presented “MIME: Adapting a Single Neural Network for Multi-task Inference with Memory-efficient Dynamic Pruning” as part of the Machine Learning for Resource Management: From Edge to Cloud session.
Brian Crafton, Zishen Wan, Samuel Spetalnick, Jong-Hyeok Yoon, and Arijit Raychowdhury of Georgia Institute of Technology, along with Intel collaborators Wei Wu, Carlos Tokunaga, and Vivek De presented “Improving Compute In-Memory ECC Reliability with Successive Correction” in the Future Unleashed: Beyond-CMOS Meets the Real World session.