INESC-ID

About: INESC-ID is a nonprofit organization based out in Lisbon, Portugal. It is known for research contribution in the topics: Field-programmable gate array & Control theory. The organization has 932 authors who have published 2618 publications receiving 37658 citations.

MITRA: byzantine fault-tolerant middleware for transaction processing on replicated databases

Abstract: Replication is often considered a cost-effective solution for building dependable systems with off-the-shelf hardware. Replication software is usually designed to tolerate crash faults, but Byzantine (or arbitrary) faults such as software bugs are well-known to affect transactional database management systems (DBMSs) as many other classes of software. Despite the maturity of replication technology, Byzantine fault-tolerant replication of databases remains a challenging problem. The paper presents MITRA, a middleware for replicating DBMSs and making them tolerant to Byzantine faults. MITRA is designed to offer transparent replication of off-theshelf DBMSs with replicas from different vendors.

Run-Time Machine Learning for HEVC/H.265 Fast Partitioning Decision

Abstract: A novel fast Coding Tree Unit partitioning for HEVC/H.265 encoder is proposed in this paper. This method relies on run-time trained neural networks for fast Coding Units splitting decisions. Contrasting to state-of-the-art solutions, this method does not require any pre-training and provides a high adaptivity to the dynamic changes in video contents. By an efficient sampling strategy and a multi-thread implementation, the presented technique successfully mitigates the computational overhead inherent to the training process on both the overall processing performance and on the initial encoding delay. The experiments show that the proposed method successfully reduces the HEVC/H.265 encoding time for up to 65% with negligible rate-distortion penalties.

Improving Speech Recognition through Automatic Selection of Age Group – Specific Acoustic Models

Abstract: The acoustic models used by automatic speech recognisers are usually trained with speech collected from young to middle-aged adults. As the characteristics of speech change with age, such acoustic models tend to perform poorly on children's and elderly people's speech. In this study, we investigate whether the automatic age group classification of speakers, together with age group -specific acoustic models, could improve automatic speech recognition performance. We train an age group classifier with an accuracy of about 95% and show that using the results of the classifier to select age group -specific acoustic models for children and the elderly leads to considerable gains in automatic speech recognition performance, as compared with using acoustic models trained with young to middle-aged adults' speech for recognising their speech, as well.