Search Results for author:
Found 9 papers, 0 papers with code
Input margins can predict generalization too
While margins have been shown to be correlated with the generalization ability of a model when measured at its hidden representations (hidden margins), no such link between large margins and generalization has been established for input margins.
The Missing Margin: How Sample Corruption Affects Distance to the Boundary in ANNs
A global estimate of margin size is usually used in the literature.
Efficient acoustic feature transformation in mismatched environments using a Guided-GAN
The GAN is used to enhance the features of mismatched data prior to decoding, or can optionally be used to fine-tune the acoustic model.
Automatic Speech Recognition
Automatic Speech Recognition (ASR)
+2
Multi-style Training for South African Call Centre Audio
Mismatched data is a challenging problem for automatic speech recognition (ASR) systems.
Automatic Speech Recognition
Automatic Speech Recognition (ASR)
+2
Exploring layerwise decision making in DNNs
While deep neural networks (DNNs) have become a standard architecture for many machine learning tasks, their internal decision-making process and general interpretability is still poorly understood.
Tracking translation invariance in CNNs
Our main focus is to investigate the effect of different architectural components of a standard CNN on that network's sensitivity to translation.
Stride and Translation Invariance in CNNs
We also observe that this characteristic is dataset-specific and dictates the relationship between pooling kernel size and stride required for translation invariance.
Pre-interpolation loss behaviour in neural networks
When training neural networks as classifiers, it is common to observe an increase in average test loss while still maintaining or improving the overall classification accuracy on the same dataset.
DNNs as Layers of Cooperating Classifiers
A robust theoretical framework that can describe and predict the generalization ability of deep neural networks (DNNs) in general circumstances remains elusive.
