A 40-310TOPS/W SRAM-Based All-Digital Up to 4b In-Memory Computing Multi-Tiled NN Accelerator in FD-SOI 18nm for Deep-Learning Edge Applications.
Giuseppe DesoliNitin ChawlaThomas BoeschManui AvodhyawasiHarsh RawatHitesh ChawlaVS AbhijithPaolo ZambottiAkhilesh SharmaCarmine CappettaMichele RossiAntonio De VitaFrancesca GirardiPublished in: ISSCC (2023)
Keyphrases
- deep learning
- dynamic random access memory
- unsupervised learning
- silicon on insulator
- unsupervised feature learning
- random access memory
- neural network
- knn
- mental models
- power consumption
- deep architectures
- pattern recognition
- machine learning
- long range
- learning strategies
- cmos technology
- text mining
- supervised learning
- active learning