Together with the IEEE Circuits and Systems Society and the IEEE Electron Devices Society, IBM Research is hosting the 4th AI Compute Symposium virtually.
The IBM and IEEE sponsored Symposium will bring together dreamers, thinkers, and innovators in cutting-edge research for a two-day forum to explore AI Compute challenges and future research directions. This symposium is free of charge.
The symposium will have eminent invited speakers as well as a student poster session. Students are encouraged to submit poster abstracts during registration. Top posters will be invited to submit long-form papers to a special issue IEEE journal. A best poster will also be awarded.
This event is an initiative of the IBM Academy of Technology.
Topics to explore
Cloud / Data Center AI Acceleration
AI at the Edge / AI in IoT
Analog Computing and Emerging Devices for AI
Neuromorphic Architectures
In Memory Computing
AI Algorithms & Applications
Approximate Computing
Speakers
Distinguished invited speakers (accepted so far)
- Venkat Thanvantri, Cadence
- Evgeni Gousev, Qualcomm
- Song Han, MIT
- Teo Laino, IBM
- Philip Wong, Stanford
- Hoi-Jun Yoo, KAIST
- Pradeep Dubey, Intel
- Steve Pawlowski, Micron
- Gunnar Hellekson, Red Hat
- SukHwan Lim, Samsung Electronics
- Tamar Eilam, IBM
Program
Submissions
Poster abstracts (2 pages maximum) are invited that address advanced results in AI Compute topics. Authors of the accepted posters will be requested to produce a short video describing the poster.
Abstract submission: 21 September 2021
Notification: 28 September 2021
Symposium: 13-14 October 2021
Winners of the 2021 Poster Sessions
Poster Track 1 - AI Architectures and Algorithms
- Rabin Acharya (University of Florida)
InfoNEAT: Information theoretic-based Neuroevolution of Augmenting Topologies for Developing Compact and Quantized Neural Networks - Wojciech Romaszkan (University of California, Los Angeles )
Precision-Tunable All-Digital Stochastic Computing Neural Network Inference Accelerator - Kevin Herisse (IEMN - Junia)
Mixed-Signal In-Memory Multi-bit Matrix-Vector Multiplication
Poster Track 2 - AI Systems and Applications
- Ankita Paul (Drexel University)
Energy Efficient Detection of Respiratory Anomaly using Spiking Neural Networks - Shamir Khandaker (University of Louisiana at Lafayette )
Automated Reduction of Irrelevant Features of a Text Classifier - Madeleine Abernot (LIRMM, University of Montpellier, CNRS )
Mobile Robot Obstacle Avoidance with Oscillatory Neural Networks on FPGA
Poster Track 3 - Emerging Hardware for AI Compute
- Murat Onen (MIT)
CMOS-compatible Protonic Programmable Resistor based on Phosphosilicate Glass Electrolyte for Analog Deep Learning - Rubab Ume (SUNY Polytechnic Institute-Albany)
Te-free Group III-Sb Binary Alloys Single Layer Test Cells for Multilevel PCM Cells - Adithi Krishnaprasad (University of Central Florida)
Implementation of Boolean logic using ultra-low variability MoS2 synapses and MoS2 LIF neurons
Posters
1 - AI Architectures and Algorithms
- DFSynthesizer: Dataflow-based Synthesis of Spiking Neural Networks to Neuromorphic Hardware
- Program-to-Circuit: Exploiting GNNs for Program Representation and Circuit Translation
- Efficient Neural Architecture Search
- Hybrid In-Memory Computing Architecture for Robust DNN Acceleration
- Approximate Computing Based Tensor Processing Unit For High-Performance AI Compute
- Characterizing Neuro-Symbolic Workloads
- Union: A Unified HW-SW Co-Design Ecosystem in MLIR for Evaluating Tensor Operations on Spatial Accelerators
- A Case for Down-Sampled Burn-In for MCMC Accelerators
- Activation Compression of DNNs through Dynamic Channel Clustering
- Silicon Photonic Architecture for Training Deep Neural Networks with Direct Feedback Alignment
- Precision-Tunable All-Digital Stochastic Computing Neural Network Inference Accelerator
- InfoNEAT: Information theoretic-based Neuroevolution of Augmenting Topologies for Developing Compact and Quantized Neural Networks
- Mixed-Signal In-Memory Multi-bit Matrix-Vector Multiplication
- Light in Artificial Intelligence: Efficient Neurocomputing with Optical Neural Networks
- An In-Situ Sliding Window Approximate Inner-Product Scheme Based on Distributed Arithmetic for Ultra-Low Power Fault-Tolerant Applications
- AI-RISC: Extending RISC-V with tightly integrated accelerators and custom instructions for AI inference at the Edge of IoT
- Analysis of Deep Spiking Neural Networks
- Bit-serial Weight Pools: Compression and Arbitrary Precision Execution of Neural Networks on Resource-Constrained Processors
- Hardware-Aware Neural Architecture Search
2 - AI Systems and Applications
- Energy Efficient Detection of Respiratory Anomaly using Spiking Neural Networks
- Real-time sepsis prediction using fusion of on-chip analog classifier and electronic medical record
- Performance Analysis of Multi-Kernel CNN Applications on CPU-FPGA Platforms with HBM
- STUDY OF THE CLIMATE ELEMENTS INFLUENCE ON LOW COST AIR QUALITY MONITORING SYSTEMS
- Multi-UAV based Data Collection and Relay to Cloud via Intelligent Edge device
- Efficient Training and Hardware Co-Design for Machine Learning models in FPGAs
- Enhancing Neural Network Performance by Using Shared Weights via Quaternions and Vector Maps
- ProxyVM: A Scalable and Retargetable CompilerFramework for Privacy-Aware Proxy Workload Generation
- Comparing AI and Entropy Techniques to Mitigate DDoS Attacks in Software Defined Networks
- A Survey on the Optimization of Deep Neural Network Accelerators for Micro-AI On-Device Inference
- Mobile Robot Obstacle Avoidance with Oscillatory Neural Networks on FPGA
- COVID-Matter: A Scalable Multimodal Sensory Machine Learning Framework for Severity Detection of Respiratory Diseases and Pandemic Prevention
- Deploying Adversarially Robust Computer Vision Deep Learning Models Across the Computing Spectrum
- Deploying AI Applications to an Unsafe Edge
- A Hybrid Capsule Network-based Deep Learning Architecture for Deciphering Ancient Scripts with Scarce Annotations
- The Need for Attention in Music Tasks
- Automated System Design and Management with Inflight Analytics
- Automated Reduction of Irrelevant Features of a Text Classifier
- Open - Source Toolchain Optimization
3 - Emerging Hardware for AI Compute
- Improving Inference Lifetime of Neuromorphic Systems via Intelligent Synapse Mapping
- Dual Gate-Tunable Memtransistor Crossbars for Higher-Order Deep Learning
- CMOS-compatible Protonic Programmable Resistor based on Phosphosilicate Glass Electrolyte for Analog Deep Learning
- Resistive Switching and Conduction Mechanisms in Ferroelectric Synaptic Weights
- Exploring Spiking Neural Network Dynamics with Memristors and R(t) Elements
- BRIM: Bistable Resistively-coupled Ising Machine
- TDM Based Memristive Triplet-STDP.
- A Low Power AdEx I&F Neuron Implementation
- Implementation of Boolean logic using ultra-low variability MoS2 synapses and MoS2 LIF neurons
- Ultra-low cycle-to-cycle variability in Au/MoS2/Ti/Au memristive synapses for neuromorphic computing
- Stuck-at-Fault Tolerance Improvement in RRAM-based DNN
- Cryogenic In-MRAM Computing Using 77K Compact Model
- Proposal of Analog In-Memory Computing with Magnified Tunnel Magnetoresistance Ratio and Universal STT MRAM Cell
- A 65nm Compute-In-Memory 7T SRAM Macro Supporting 4-bit Multiply and Accumulate (MAC) Operation by Employing Charge Domain Compute
- Reaching Autonomy at Insect-Scale: A Compute-in-Memory Pathway for Robustness and Ultra-low-power
- Hybrid NN Model to Predict the Radiation Damage in GaN HEMTs
- Te-free Group III-Sb Binary Alloys Single Layer Test Cells for Multilevel PCM Cells
Organizers
Advisory Committee
- Amara Amara (IEEE CAS)
- Fernando Guarin (GF)
- Hai Li (Duke University)
- Yong Lion (IEEE CAS)
- Ravi Todi (IEEE EDS)
- IBM Academy of Technology
- John Handy Bosma (IBM)
Poster Session Chairs
- Krishnan Kailas
- Jin-Ping Han
- Kaoutar El Maghraoui
General Chair
- Rajiv Joshi
Program Committee
- Matthew M. Ziegler
- Arvind Kumar
- Kaoutar El Maghraoui
- John Rozen
- Krishnan Kailas
- Jin-Ping Han
- Xin Zhang
- Anna Topol
Publicity Chairs
Registration Chair
ThinkLab/Auditorium Event Head
- Jenna Gray
- George Tulevski
Audio/Video
- Jim Fuhs (Fusion Marketing)
IBM Web Support
(IBM Research Europe)
- Linda Rudin (Zurich, Switzerland)