Figure 4 | Scientific Reports

Figure 4

From: Stochastic binary synapses having sigmoidal cumulative distribution functions for unsupervised learning with spike timing-dependent plasticity

Figure 4

Recognition accuracy. (a) Evolutions of recognition accuracy obtained using sigmoidal stochastic S-STDP with the number of learnings for \(k=3\) (with \(p=0.13\) and \(q=0.03\)) and \(4\) (with \(p=0.2\) and \(q=0.08\)) as well as those using conventional S-STDP (\(k=1\) with \(p=0.04\) and \(q=0.008\)) and deterministic S-STDP. The horizontal broken line indicates the 90% accuracy level. (b) Comparison of the accuracies achieved in this work with previously reported values34,44. (c) Relationship between the recognition accuracy and the number of neurons retaining their initially memorised digits after training with 150,000 samples. Data are obtained with various combinations of \(k\), \(p\) and \(q\). The broken line is a visual guide.

Back to article page