INESC-ID

About: INESC-ID is a nonprofit organization based out in Lisbon, Portugal. It is known for research contribution in the topics: Computer science & Context (language use). The organization has 932 authors who have published 2618 publications receiving 37658 citations.

{2/sup n/ + 1, 2/sup n+k/, 2/sup n/ - 1} : a new RNS moduli set extension

Abstract: The increasing usage of residual number system (RNS) in signal processing applications demands the development of new and more adaptable RNS moduli sets and arithmetic units. This paper presents a new adaptable moduli set extension for the traditional moduli set {2/sup n/ + 1, 2/sup n/, 2/sup n/ - 1}. As it will be shown, this new moduli set extension ({2/sup n/ + 1, 2/sup n+k/, 2/sup n/ - 1}) allows the balancing of the binary channel (2/sup n+k/) in relation to the other two channels. Moreover, it does not require the development of new addition and multiplication units, since it is possible to reuse the already developed and well studied units for these moduli operations.

Bilateral negotiation in energy markets: Strategies for promoting demand response

Abstract: Two major goals of electricity markets are ensuring a secure and efficient operation and decreasing the cost of energy. To achieve these goals, three major market models have been considered: pools, bilateral contracts and hybrid markets. Pool prices tend to change quickly and variations are usually highly unpredictable. In this way, market participants can enter into bilateral contracts to hedge against pool price volatility.Multi-agent electricity markets-that is, energy management tools based on software agents-have received some attention lately and a number of prominent simulators have been proposed in the literature. However, despite the power and elegance of existing tools, they often lack generality and flexibility, mainly because they are limited to particular features of market players. This paper describes on-going work that uses the potential of agent-based technology to develop a computational tool to support bilateral contracting in electricity markets. Specifically, the purpose of the paper is threefold: (i) to present the key features of a model for software agents that handles twoparty and multi-issue negotiation, (ii) to describe two novel negotiation strategies for promoting demand response, a "volume management" strategy for end-use consumers, and a "price management" strategy for producers/retailers, and (iii) to describe a case study on forward bilateral contracts, involving a retailer agent and a commercial customer.

Multi-core for K-means clustering on FPGA

Abstract: In this paper, a configurable many-core hardware/software architecture is proposed to efficiently execute the widely known and commonly used K-means clustering algorithm. A prototype was designed and implemented on a Xilinx Zynq-7000 All Programmable SoC. A single core configured with the slowest configuration achieves a 10× speed-up compared to the software only solution. The system is fully scalable and capable of achieving much higher speed-ups by increasing its parallelism.

Twitter gender classification using user unstructured information

Abstract: This paper describes an approach to automatically detect the gender of Twitter users, based only on clues provided by their profile information in an unstructured form. A number of features that capture phenomena specific of Twitter users is proposed and evaluated on a dataset of about 242K English language users. Different supervised and unsupervised approaches are used to assess the performance of the proposed features, including Naive Bayes variants, Logistic Regression, Support Vector Machines, Fuzzy c-Means clustering, and K-means. An unsupervised approach based on Fuzzy c-Means proved to be very suitable for this task, returning the correct gender for about 96% of the users.

Back-Translation-Style Data Augmentation for End-to-End ASR

Abstract: In this paper we propose a novel data augmentation method for attention-based end-to-end automatic speech recognition (E2E-ASR), utilizing a large amount of text which is not paired with speech signals. Inspired by the back-translation technique proposed in the field of machine translation, we build a neural text-to-encoder model which predicts a sequence of hidden states extracted by a pre-trained E2E-ASR encoder from a sequence of characters. By using hidden states as a target instead of acoustic features, it is possible to achieve faster attention learning and reduce computational cost, thanks to sub-sampling in E2E-ASR encoder, also the use of the hidden states can avoid to model speaker dependencies unlike acoustic features. After training, the text-to-encoder model generates the hidden states from a large amount of unpaired text, then E2E-ASR decoder is retrained using the generated hidden states as additional training data. Experimental evaluation using LibriSpeech dataset demonstrates that our proposed method achieves improvement of ASR performance and reduces the number of unknown words without the need for paired data.