Automatic Node Selection for Deep Neural Networks using Group Lasso Regularization
This work addresses node selection for efficiency in speech recognition systems, but it is incremental as it builds on existing regularization methods.
The study tackled the problem of selecting salient nodes in deep neural networks for speech recognition by applying Group Lasso regularization to a DNN-HMM hybrid on TED Talks data, resulting in successful node selection that maintained high classification power.
We examine the effect of the Group Lasso (gLasso) regularizer in selecting the salient nodes of Deep Neural Network (DNN) hidden layers by applying a DNN-HMM hybrid speech recognizer to TED Talks speech data. We test two types of gLasso regularization, one for outgoing weight vectors and another for incoming weight vectors, as well as two sizes of DNNs: 2048 hidden layer nodes and 4096 nodes. Furthermore, we compare gLasso and L2 regularizers. Our experiment results demonstrate that our DNN training, in which the gLasso regularizer was embedded, successfully selected the hidden layer nodes that are necessary and sufficient for achieving high classification power.