Massively Deep Artificial Neural Networks for Handwritten Digit Recognition
arXiv:1507.05053v13 citations
Originality Synthesis-oriented
AI Analysis
This is an incremental improvement for digit recognition tasks, as it builds on existing methods with deeper architectures and hardware optimization.
The paper tackled handwritten digit recognition on the MNIST database, achieving a 0.72% error rate using a deep neural network with many hidden layers and GPU acceleration.
Greedy Restrictive Boltzmann Machines yield an fairly low 0.72% error rate on the famous MNIST database of handwritten digits. All that was required to achieve this result was a high number of hidden layers consisting of many neurons, and a graphics card to greatly speed up the rate of learning.