# LocalTrain: Integer-Only Local Learning for Energy-Efficient CNN Training on Edge FPGAs

arXiv: https://arxiv.org/abs/2610.17099 · HTML: https://mlsanity.com/paper/2610.17099/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 12 references that are in the map.
- First posted on arXiv 12 Oct 2026.

## Builds on (most cited first, 10 of 10)

- [The Forward-Forward Algorithm: Some Preliminary Investigations](https://mlsanity.com/paper/2212.13345/) (2022)
- [Training Neural Networks with Local Error Signals](https://mlsanity.com/paper/1901.06656/) (2019)
- [Greedy Layerwise Learning Can Scale to ImageNet](https://mlsanity.com/paper/1812.11446/) (2018)
- [Flexpoint: An Adaptive Numerical Format for Efficient Training of Deep Neural Networks](https://mlsanity.com/paper/1711.02213/) (2017)
- [Convolutional Channel-wise Competitive Learning for the Forward-Forward Algorithm](https://mlsanity.com/paper/2312.12668/) (2023)
- [NITI: Training Integer Neural Networks Using Integer-only Arithmetic](https://mlsanity.com/paper/2009.13108/) (2020)
- [EF-Train: Enable Efficient On-device CNN Training on FPGA Through Data Reshaping for Online Adaptation or Personalization](https://mlsanity.com/paper/2202.10935/) (2022)
- [Automatic Compiler Based FPGA Accelerator for CNN Training](https://mlsanity.com/paper/1908.06724/) (2019)
- [FF-INT8: Efficient Forward-Forward DNN Training on Edge Devices with INT8 Precision](https://mlsanity.com/paper/2506.22771/) (2025)
- [NITRO-D: Native Integer-only Training of Deep Convolutional Neural Networks](https://mlsanity.com/paper/2407.11698/) (2024)

## Abstract (arXiv)

Local learning reduces cross-layer training dependencies, but an efficient accelerator must still retain operands, accumulate gradients, and select quantization scales. We present an FPGA system that trains a CNN and its regional auxiliary classifiers using a shared integer matrix engine. Compact errors are expanded on chip, gradients are folded tile by tile, and historical scales permit direct output conversion. At 200 MHz, the KV260 prototype achieves 76.57% accuracy on CIFAR-10 and 91.85% on SVHN at approximately 4.20 W mean module power. Effective matrix throughput is 31.22 and 31.57 GMAC/s, respectively. Against the same integer method on Jetson Orin Nano, measured training energy is 40.4% and 42.1% lower, with 1.86× and 1.82× longer runtime; measurements use the respective platform power domains. Selected state checks and matching complete-run prediction counts support functional agreement with the software reference.