ELoNC
Project: Exploring the Logic of Neural Computing
Collaborating departments: Electrical and Computer Engineering (TUM); Computer and Communication Sciences (EPFL)
From neural networks to circuits
Inspired by the human brain, neural networks attempt to emulate the way the brain works. The last decade has witnessed the breakthrough of neural networks in various fields, e.g., image/speech processing. Despite of the huge success of neural networks, their enormous power consumption executing on digital hardware, e.g., TPU and GPU, restricts their applications in resource-constrained platforms, e.g., edge devices. The large power consumption results from a huge number of MAC operations and data movement.
To address the power consumption problem, ELoNC, a joint project of TUM and EPFL targets to explore the relationship between neural networks and their circuit implementations. Specifically, the feasibility of directly converting neural networks into digital circuits will be verified. The team at TUM will focus on developing various training methods to enable the conversion of neural networks into circuits and enhance logic synthesis for such conversion. The team at EPFL will support the logic synthesis and verify the resulting circuits on FPGAs.
Kangwei Xu, Grace Li Zhang, Xunzhao Yin, Cheng Zhuo, Ulf Schlichtmann, Bing Li, “Automated C/C++ Program Repair for High-Level Synthesis via Large Language Models,” ACM/IEEE International Symposium on Machine Learning for CAD, 2024
Ruidi Qiu, Grace Li Zhang, Rolf Drechsler, Ulf Schlichtmann, Bing Li, “AutoBench: Automatic Testbench Generation and Evaluation Using LLMs for HDL Design,” ACM/IEEE International Symposium on Machine Learning for CAD, 2024
Ruidi Qiu, Amro Eldebiky, Grace Li Zhang, Xunzhao Yin, Cheng Zhuo, Ulf Schlichtmann, Bing Li, “OplixNet: Towards Area-Efficient Optical Split-Complex Networks with Real-to-Complex Data Assignment and Knowledge Distillation,” Design, Automation and Test in Europe (DATE), 2024
Kangwei Xu, Grace Li Zhang, Ulf Schlichtmann, Bing Li, “Logic Design of Neural Networks for High-Throughput and Low-Power Applications,” IEEE/ACM Asia and South Pacific Design Automation Conference (ASP-DAC), 2024
Wenhao Sun, Grace Li Zhang, Huaxi Gu, Bing Li, Ulf Schlichtmann, “Class-based Quantization for Neural Networks,” Design, Automation and Test in Europe (DATE), 2023
Wenhao Sun, Grace Li Zhang, Xunzhao Yin, Cheng Zhuo, Huaxi Gu, Bing Li, Ulf Schlichtmann, “SteppingNet: A Stepping Neural Network with Incremental Accuracy Enhancement,” Design, Automation and Test in Europe (DATE), 2023
Richard Petri, Grace Li Zhang, Yiran Chen, Ulf Schlichtmann, Bing Li, “PowerPruning: Selecting Weights and Activations for Power-Efficient Neural Network Acceleration,” ACM/IEEE Design Automation Conference (DAC), 2023
Team
Coordinating Postdoc
PD Dr.-Ing. habil. Bing Li
Chair of Electronic Design Automation | TUM
Doctoral Candidate
Riudi Qui
Chair of Electronic Design Automation | TUM
Doctoral Candidate
Wenhao Sun
Chair of Electronic Design Automation | TUM
Principal Investigator
Prof. Dr.-Ing. Ulf Schlichtmann
Chair of Electronic Design Automation | TUM
Principal Investigator
Professor Paolo Ienne
Processor Architecture Laboratory | EPFL