Exposing numerical bugs in deep learning via gradient back-propagation.
Ming YanJunjie ChenXiangyu ZhangLin TanGan WangZan WangPublished in: ESEC/SIGSOFT FSE (2021)
Keyphrases
- back propagation
- deep learning
- artificial neural networks
- neural network
- feed forward
- bp algorithm
- training algorithm
- feed forward neural networks
- steepest descent method
- unsupervised learning
- neural nets
- bp neural network
- learning algorithm
- multilayer perceptron
- unsupervised feature learning
- fuzzy logic
- mental models
- hidden layer
- weakly supervised
- machine learning
- weight update
- artificial intelligence
- genetic algorithm