Close

Session

Research Manuscript
:
Design for the Unpredictable: Resilience and Adaptation in AIoT
DescriptionThis session presents recent advances in the design of components and methodologies that enable resilient, reliable, and low-energy systems. The presentations cover error-correction codes for neural networks, the design of safe and reliable NPUs and controllers, run-time analyzers, and innovative techniques for reasoning and adaptation. Together, these talks demonstrate how cross-layer hardware components, design methodologies, and run-time analysis are shaping the next generation of robust and trustworthy autonomous systems.
Event Type
Research Manuscript
TimeTuesday, July 2810:30am - 12:30pm PDT
LocationMtg Room 202C
Topics
Systems
Tracks
SYS2. Design of Cyber-Physical Systems and IoT
Presentations
10:30am - 10:43am PDTE³-CODE: Embedded and Efficient Error-Correcting Code for Error-Resilient Neural Networks
10:43am - 10:56am PDTFrom Characterization to Microarchitecture: Designing an Elegant and Reliable BFP-Based NPU
10:56am - 11:10am PDTProvably Probabilistic Safe Controller Synthesis for Vision-Based Neural Network Control Systems
11:10am - 11:23am PDTWINNER: A Wireless In-Vivo/ex-Vivo Runtime Analyzer for Intermittent Computing
11:23am - 11:36am PDTEnergyhdc: Intermittent Hyperdimensional Inference on Energy-Harvesting IoT
11:36am - 11:50am PDTLLM-Enhanced Two-Stage Signal Temporal Logic Automatic Transformation with Structured Intermediate Representation
11:50am - 12:03pm PDTTorr: Towards Brain-Inspired Task-Oriented Reasoning via Cache-Oriented Algorithm-Architecture Co-Design
12:03pm - 12:16pm PDTHyperlidar: Adaptive Post-Deployment LiDAR Segmentation via Hyperdimensional Computing
12:16pm - 12:30pm PDTRTFL: Energy-Aware Federated Learning for AIoT Design Via Adaptive Quantization-Based Multi-Agent Scheduling