Name
Abstract | ||
|---|---|---|
Reduced-precision arithmetic improves the size, cost, power and performance of neural networks in digital logic. In convolutional neural networks, the use of 1b weights can achieve state-of-the-art error rates while eliminating multiplication, reducing storage and improving power efficiency. The BinaryConnect binary-weighted system, for example, achieves 9.9% error using floating-point activations on the CIFAR-10 dataset. In this paper, we introduce TinBiNN, a lightweight vector processor overlay for accelerating inference computations with 1b weights and 8b activations. The overlay is very small -- it uses about 5,000 4-input LUTs and fits into a low cost iCE40 UltraPlus FPGA from Lattice Semiconductor. To show this can be useful, we build two embedded u0027person detectoru0027 systems by shrinking the original BinaryConnect network. The first is a 10-category classifier with a 89% smaller network that runs in 1,315ms and achieves 13.6% error. The other is a 1-category classifier that is even smaller, runs in 195ms, and has only 0.4% error. In both classifiers, the error can be attributed entirely to training and not reduced precision. |
Year | Venue | DocType |
|---|---|---|
2019 | arXiv: Distributed, Parallel, and Cluster Computing | Journal |
Volume | Citations | PageRank |
abs/1903.06630 | 0 | 0.34 |
References | Authors | |
0 | 9 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Guy G. F. Lemieux | 1 | 179 | 14.96 |
| Joe Edwards | 2 | 0 | 0.68 |
| Joel Vandergriendt | 3 | 0 | 0.34 |
| Aaron Severance | 4 | 64 | 5.19 |
| Ryan De Iaco | 5 | 0 | 0.34 |
| Abdullah Raouf | 6 | 0 | 0.34 |
| Hussein Osman | 7 | 0 | 0.34 |
| Tom Watzka | 8 | 0 | 0.34 |
| Satwant Singh | 9 | 0 | 0.34 |