Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

Digital processing of speech signals

An intelligent automated assistant system engages with the user in an integrated, conversational manner using natural language dialog, and invokes external services when appropriate to obtain information or perform various actions. The system can be implemented using any of a number of different platforms, such as the web, email, smartphone, and the like, or any combination thereof. In one embodiment, the system is based on sets of interrelated domains and tasks, and employs additional functionally powered by external services with which the system can interact.

Intelligent Automated Assistant

When building a unified vision system or gradually adding new capabilities to a system, the usual assumption is that training data for all tasks is always available. However, as the number of tasks grows, storing and retraining on such data becomes infeasible. A new problem arises where we add new capabilities to a Convolutional Neural Network (CNN), but the training data for its existing capabilities are unavailable. We propose our Learning without Forgetting method, which uses only new task data to train the network while preserving the original capabilities. Our method performs favorably compared to commonly used feature extraction and fine-tuning adaption techniques and performs similarly to multitask learning that uses original task data we assume unavailable. A more surprising observation is that Learning without Forgetting may be able to replace fine-tuning with similar old and new task datasets for improved new task performance.

Learning without Forgetting

The method involves determining partial deviations for each of multiple successive frames of time lengths of a reference speech signal (6) and measuring speech signal (16) in a data processing unit (18) as a function of phonetic reference information (8). Total deviation is determined from the partial deviations and the phonetic reference information assigned to the respective frame. The partial deviations are weighted by a weighting factor, which is dependent on the assigned phonetic reference information such that a determined deviation is noticed in transmitted signals. An independent claim is also included for a device for measuring quality of speech signal.

Method for measuring quality of speech signal transmitted through e.g. voice over internet protocol, involves weighing partial deviations of each frames of time lengths of reference, and measuring speech signals by weighting factor

The emergence of data-driven machine learning (ML) has facilitated significant progress in many complicated tasks such as highly-automated driving. While much effort is put into improving the ML models and learning algorithms in such applications, little focus is put into how the training data and/or validation setting should be designed. In this paper we investigate the influence of several data design choices regarding training and validation of deep driving models trainable in an end-to-end fashion. Specifically, (i) we investigate how the amount of training data influences the final driving performance, and which performance limitations are induced through currently used mechanisms to generate training data. (ii) Further, we show by correlation analysis, which validation design enables the driving performance measured during validation to generalize well to unknown test environments. (iii) Finally, we investigate the effect of random seeding and non-determinism, giving insights which reported improvements can be deemed significant. Our evaluations using the popular CARLA simulator provide recommendations regarding data generation and driving route selection for an efficient future development of end-to-end driving models.

On the Choice of Data for Efficient Training and Validation of End-to-End Driving Models

Investigations on Offline Artificial Bandwidth Extension of Telephone Speech Databases.

The obligation to press a push-to-speak button before issuing a voice command to a speech dialog system is not only inconvenient but it also leads to decreased recognition accuracy if the user starts speaking prematurely. In this chapter, we investigate the performance of a so-called talk-and-push (TAP) system, which permits the user to begin an utterance within a certain time frame before or after pressing the button. This is achieved using a speech signal buffer in conjunction with an acoustic echo cancelation unit and a combined noise reduction and start-of-utterance detection. In comparison with a state-of-the-art system employing loudspeaker muting, the TAP system delivers significant improvements in the word error rate.

Tim Fingscheidt

Papers

Method for measuring quality of speech signal transmitted through e.g. voice over internet protocol, involves weighing partial deviations of each frames of time lengths of reference, and measuring speech signals by weighting factor

On the Choice of Data for Efficient Training and Validation of End-to-End Driving Models

Investigations on Offline Artificial Bandwidth Extension of Telephone Speech Databases.

A Novel Way to Start Speech Dialogs in Cars by Talk-and-Push (TAP)

Implementation aspects of the GSM half-rate speech codec.