November 16th, 2022 (Wednesday)
Location: ECE 202, New Jersey Institute of Technology, Newark, NJ, USA
|
Location: ECE 202, New Jersey Institute of Technology, Newark, NJ, USA |
|
Future computing systems are expected to exploit our improved understanding of the brain through leveraging similar computational principles. In this way, Neuromorphic computing has shown the potential for breakthroughs in machine intelligence. The main goal of this tutorial is to dive deep into the rapidly developing field of Neuromorphic Computing and AI and cover its cross-layer design challenges from device to algorithms. The IEEE EDS Tutorial on Neuromorphic Computing (TNC 2022) offers three talks and tutorials by leading researchers from multiple disciplines and prominent universities and promotes student presentations (posters) to demonstrate new research and results, discuss the potential and challenges of the Neuromorphic Computing Devices, future research needs, and directions, and shape collaborations.
|
Call for Posters Postdoc, Ph.D./M.Sc./U.G. students are encouraged to register and submit a title and a 100-word abstract to present their latest research results as a poster at the event. Registration Link |
|
Program Schedule (Times in EST) |
| 9:00 AM - 9:30 AM | Breakfast, Coffee, Registration, Networking, and Poster Set up |
| 9:30 AM - 9:50 AM | Poster Presentation |
|
9:50 AM - 10:00 AM |
Opening Remarks by NJIT ECEChair Dr. Misra |
|
10:00 AM - 10:50 AM
|
Talk I: Dr. Jack (Qingxue) Zhang (Purdue University) Title: Brain-Inspired AI on the Edge for Precision Medicine Abstract: Deep learning-empowered precision medicine is of great promise for mining insights from the human data. If deployed on the edge devices like smart phones and health monitors, the deep learning models will be able to provide real-time inference ability for instantaneous data analytics and medical decision support. Dr. Zhang’s lab is uniquely combing AI theories, neuromorphic learning algorithms, efficient computing, and wearable/IoT monitoring, targeting the smart health and world applications. This talk will highlight the efforts on both efficient learning and neuromorphic learning for wearable big data applications. Starting with an overview of the field, state-of-the-art, challenges, and opportunities, Dr. Zhang will then highlight two selected projects: knowledge distillation learning for edge-deployable AI, and backward neuromorphic learning algorithms for spiking neural networks, for smart health applications. For the former one, the knowledge distillation principle that has reduced the model complexity by around 50 times will be introduced. For the latter one, an algorithm, SpikeBASE (Spiking neural network learning with Backward Adaptation of Synaptic Efflux) will be introduced, which has achieved comparable performance as standard deep learning. Several other projects will be also introduced in brief during the talk. With great thanks to ECE at New Jersey Institute of Technology, Dr. Zhang is very looking forward to the talk and communicating with experts in various fields. |
|
10:50 AM - 11:40 AM
|
Talk II: Dr. Bo Yuan (Rutgers University) Title: Algorithm and Hardware Co-Design for Efficient Deep Learning: Sparse and Low-rank Perspective Abstract: In the emerging artificial intelligence era, deep neural networks (DNNs), a.k.a. deep learning, have gained unprecedented success in various applications. However, DNNs are usually storage intensive, computation intensive and very energy consuming, thereby posing severe challenges on the future wide deployment in many application scenarios, especially for the resource-constraint low-power IoT application and embedded systems. In this talk, I will introduce the algorithm/hardware co-design works for energy-efficient DNN in my group, from both the sparse and low-rank perspectives. First, I will show the benefit of using structured and unstructured sparsity of DNN for designing low-latency and low-power DNN hardware accelerators. In the second part of my talk, I will present an algorithm/hardware co-design framework that leverages low tensor rankness towards energy-efficient high-accuracy DNN model and accelerators. |
|
11:40 AM - 12:30 PM
|
Talk III: Dr. Shaahin Angizi (New Jersey Institute of Technology) Title: Toward Opportunistic and Fast Integrated Sensing and Computing for Edge Imaging Systems Abstract: Internet of Things (IoT) devices are projected to attain an $1100B market by 2025, with a web of interconnection projected to comprise approximately 75+ billion IoT devices. The large number of IoTs consist of sensory imaging systems that enable massive data collection from the environment and people. However, considerable portions of the captured sensory data are redundant and unstructured. Data conversion of such large raw data, storing in volatile memories, transmission, and computation in on-/off-chip processors, impose high energy consumption, latency, and a memory bottleneck at the edge. In this talk, I will be focusing on cross-layer (circuit/architecture/application) co-design of energy-efficient and high-performance processing-in-sensor and processing-in-memory platforms. This enables a smooth transition from the current cloud-centric IoT approach to a data-centric approach, whereby the mobile edge devices can opportunistically perform computation close to the sensor by repurposing the sensor/cache to a data-parallel processing unit remarkably reducing the power consumption and latency of data transmission to a back-end processor. Moreover, I will show how our research systematically enables deploying new foundational approximate, hardware-oriented, and multiply-accumulate (MAC)-free AI algorithms into resource-constrained edge devices for efficient feature extraction, reducing the computation complexity and memory access while maintaining accuracy in imaging systems. |
|
12:30 PM - 1:30 PM |
Lunch & Poster Presentation |
| 1:30 PM | Conclusion |
Organizers |
Dr. Durga Misra (IEEE Fellow), Dr. Shaahin Angizi (IEEE Member) |
