aiqbench analysis
Chameleon addresses the critical hardware bottleneck in edge AI by unifying learning and inference architectures to support few-shot and continual learning on sequential data with minimal area overhead. By utilizing a multiplier-free compute array optimized for temporal convolutional networks, it achieves state-of-the-art accuracy records while maintaining extreme power efficiency suitable for raw audio processing at 16-kHz sampling rates. This work demonstrates that specialized hardware can effectively bridge the gap between theoretical machine learning capabilities and practical deployment constraints on resource-constrained devices.
Chameleon: A Multiplier-Free Temporal Convolutional Network Accelerator for End-to-End Few-Shot and Continual Learning from Sequential Data
arXiv:2505.24852v3 Announce Type: replace-cross Abstract: On-device learning at the edge enables low-latency, private personalization with improved long-term robustness and reduced maintenance costs. Yet, achieving scalable, low-power end-to-end on-chip learning, especially from real-world sequential data with a limited number of examples, is an open challenge. Indeed, accelerators supporting error backpropagation optimize for learning performance at the expense of inference efficiency, while simplified learning algorithms often fail to reach acceptable accuracy targets. In this work, we present Chameleon, leveraging three key contributions to solve these challenges. (i) A unified learning and inference architecture supports few-shot learning (FSL), continual learning (CL) and inference at only 0.5% area overhead to the inference logic. (ii) Long temporal dependencies are efficiently captured with temporal convolutional networks (TCNs), enabling the first demonstration of end-to-end on-chip FSL and CL on sequential data and inference on 16-kHz raw audio. (iii) A dual-mode, multiplier-free compute array allows either matching the power consumption of state-of-the-art inference-only keyword spotting (KWS) accelerators or enabling $4.3\times$ higher peak GOPS. Fabricated in 40-nm CMOS, Chameleon sets new accuracy records on Omniglot for end-to-end on-chip FSL (96.8%, 5-way 1-shot, 98.8%, 5-way 5-shot) and CL (82.2% final accuracy for learning 250 classes with 10 shots), while maintaining an inference accuracy of 93.3% on the 12-class Google Speech Commands dataset at an extreme-edge power budget of 3.1 $\mu$W.
Abstract:On-device learning at the edge enables low-latency, private personalization with improved long-term robustness and reduced maintenance costs. Yet, achieving scalable, low-power end-to-end on-chip learning, especially from real-world sequential data with a limited number of examples, is an open challenge. Indeed, accelerators supporting error backpropagation optimize for learning performance at the expense of inference efficiency, while simplified learning algorithms often fail to reach acceptable accuracy targets. In this work, we present Chameleon, leveraging three key contributions to solve these challenges. (i) A unified learning and inference architecture supports few-shot learning (FSL), continual learning (CL) and inference at only 0.5% area overhead to the inference logic. (ii) Long temporal dependencies are efficiently captured with temporal convolutional networks (TCNs), enabling the first demonstration of end-to-end on-chip FSL and CL on sequential data and inference on 16-kHz raw audio. (iii) A dual-mode, multiplier-free compute array allows either matching the power consumption of state-of-the-art inference-only keyword spotting (KWS) accelerators or enabling $4.3\times$ higher peak GOPS. Fabricated in 40-nm CMOS, Chameleon sets new accuracy records on Omniglot for end-to-end on-chip FSL (96.8%, 5-way 1-shot, 98.8%, 5-way 5-shot) and CL (82.2% final accuracy for learning 250 classes with 10 shots), while maintaining an inference accuracy of 93.3% on the 12-class Google Speech Commands dataset at an extreme-edge power budget of 3.1 $\mu$W.
Hardware Architecture (cs.AR); Machine Learning (cs.LG)
C.3; B.6.0; B.7.0; I.2.6; B.5.0
arXiv:2505.24852 [cs.AR]
(or arXiv:2505.24852v3 [cs.AR] for this version)
IEEE Journal of Solid-State Circuits, vol. 61, no. 7, pp. 3814-3829, July 2026
https://doi.org/10.1109/JSSC.2025.3645640 DOI(s) linking to related resources
Full text under the publisher's open licence · 253 words
Related coverage
The current bottleneck is political will, not research
AI Alignment Forum · Jul 11, 2026
Robust Active-Perception Control for Global-State-Free Aerial-Ground Cooperation
arXiv · Sep 24, 2026
Links #5: 2026/07
LessWrong · Jul 31, 2026
Knowledge as Orbit: Finite Collections as Phases of an Exactly Periodic Latent Generator
arXiv · Sep 16, 2026