Binary neural network
Binary neural network is an artificial neural network, where commonly used floating-point weights are replaced with binary ones.[1]
It saves storage and computation, and serves as a technique for deep models on resource-limited devices. Using binary values can bring up to 58 times speedup.[2] Accuracy and information capacity of binary neural network can be manually controlled.[3] Binary neural networks do not achieve the same accuracy as their full-precision counterparts, but improvements are being made to close this gap.
It has been argued theoretically that quantum computers could train binary neural nets to their global optimum via quantum search techniques in shorter time than brute force classical optimisation.[4]
References[change | change source]
- ↑ Courbariaux, M.; Bengio, Y.; David, J.-P. (2015). "BinaryConnect: training deep neural networks with binary weights during propagation". NIPS. arXiv:1511.00363.
- ↑ Rastegari, M.; Ordonez, V.; Redmon, J.; Farhadi, A. (2016). "XNOR-Net: ImageNet classification using binary convolutional neural networks". ECCV. arXiv:1603.05279.
- ↑ Ignatov, D.; Ignatov, A. (2020). "Controlling information capacity of binary neural network". Pattern Recognition Letters. 138: 276–281. arXiv:2008.01438. Bibcode:2020PaReL.138..276I. doi:10.1016/j.patrec.2020.07.033. S2CID 220961716.
- ↑ Liao, Y.; Ebler, D.; Liu, F.; Dahlsten, O. (2021). "Quantum speed-up in global optimization of binary neural nets". New Journal of Physics. 23 (6). doi:10.1088/1367-2630/abc9ef. S2CID 228865652.