Abstract: Described herein are systems and methods to identify and address sources of bias in an end-to-end speech model. In one or more embodiments, the end-to-end model may be a recurrent neural network with two 2D-convolutional input layers, followed by multiple bidirectional recurrent layers and one fully connected layer before a softmax layer. In one or more embodiments, the network is trained end-to-end using the CTC loss function to directly predict sequences of characters from log spectrograms of audio. With optimized recurrent layers and training together with alignment information, some unwanted bias induced by using purely forward only recurrences may be removed in a deployed model.

Abstract: Described herein are systems and methods to identify and address sources of bias in an end-to-end speech model. In one or more embodiments, the end-to-end model may be a recurrent neural network with two 2D-convolutional input layers, followed by multiple bidirectional recurrent layers and one fully connected layer before a softmax layer. In one or more embodiments, the network is trained end-to-end using the CTC loss function to directly predict sequences of characters from log spectrograms of audio. With optimized recurrent layers and training together with alignment information, some unwanted bias induced by using purely forward only recurrences may be removed in a deployed model.

Abstract: A method and system are provided for online sparse regularized joint analysis for heterogeneous data. The method generates a latent space model modeling a latent space in which correlation information is encoded for a plurality of heterogeneous data points at respective time instants, responsive to respective energy-preserving projections and structure-preserving projections of the data points in the latent space. The method performs online anomaly detection on a current one of the data points responsive to the encoded correlation information for respective ones of the energy-preserving projections and structure-preserving projections for a previous one of the data points without anomaly. The method generates an alarm responsive to a detection of an anomaly for the current one of the data points. The method updates the latent space model for the current one of the data points, by a processor-based online model updater, responsive to a lack of the detection of the anomaly.