Analog cam hardware accelerates tree machine learning with noise resilience

Efficient and Variation-Tolerant Tree Inference via Compact and Scalable Mapping in Analog CAMs

Emerging Technologies

Summary

Tree-based machine learning models are good for understanding data but slow to run on normal computers because they jump around memory a lot. The authors created ACTIVAR, a new hardware-software approach that uses a special kind of memory called analog CAM to run these tree models faster and with less energy. Their method cleverly packs tree parts together to save space, trains models to handle hardware noise, and designs hardware to make decisions clearer. This lets ACTIVAR keep accuracy close to software models while being much smaller, more energy-efficient, and better at dealing with errors than previous analog CAM designs.

What this means in practice

Authors

Zhicheng Xu, Wenqing Shi, Pengyu Ren, Chao Li, Bo Wen, Ruibin Mao, Xunzhao Yin, Xiaobo Sharon Hu, Can Li

Abstract

Tree-based machine learning methods are widely adopted for their interpretability and state-of-the-art accuracy on tabular data. These methods face efficiency challenges on conventional architectures due to irregular memory access patterns during tree traversal. Prior work uses analog CAM to accelerate tree traversal, but inefficient mapping requires exponentially more analog CAM columns as tree depth increases. Accuracy is also degraded by device non-idealities such as programming variations and soft decision boundaries. To address these challenges, we propose ACTIVAR, a compact variation-tolerant tree implementation in a dense analog CAM array via software-hardware codesign. On the software side, we introduce an analog-CAM-aware compact tree model that employs hierarchical subtrees with shared features to maximize hardware utilization. This dense mapping introduces redundancy, improving both model accuracy and hardware defect tolerance. Additionally, we develop Variation-Enhanced Training, which utilizes a fuzzy overlap boundary strategy to build models resilient to analog noise. On the hardware side, we design a FeFET-based analog CAM cell with a driver to sharpen decision thresholds and implement a pipelined multi-core architecture to optimize throughput and energy consumption. Across 13 classification and regression datasets, ACTIVAR maintains near-software-level accuracy while reducing hardware overhead by up to 99% compared with prior analog-CAM-based soft-tree designs. Under realistic hardware noise conditions, ACTIVAR limits accuracy degradation to just 3.7% on the MNIST dataset, whereas baseline models suffer a catastrophic 49.6% drop. Furthermore, the ACTIVAR architecture achieves energy efficiency gains ranging from $10^4$ to $10^6$ times over commercial general-purpose processors, establishing it as a highly efficient solution for high-capacity machine learning at the edge.