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.

Graph-based Semi-Supervised Model for Joint Chinese Word Segmentation and Part-of-Speech Tagging

Abstract: This paper introduces a graph-based semisupervised joint model of Chinese word segmentation and part-of-speech tagging. The proposed approach is based on a graph-based label propagation technique. One constructs a nearest-neighbor similarity graph over all trigrams of labeled and unlabeled data for propagating syntactic information, i.e., label distributions. The derived label distributions are regarded as virtual evidences to regularize the learning of linear conditional random fields (CRFs) on unlabeled data. An inductive character-based joint model is obtained eventually. Empirical results on Chinese tree bank (CTB-7) and Microsoft Research corpora (MSR) reveal that the proposed model can yield better results than the supervised baselines and other competitive semi-supervised CRFs in this task.

Predicting progression of mild cognitive impairment to dementia using neuropsychological data: a supervised learning approach using time windows

Abstract: Predicting progression from a stage of Mild Cognitive Impairment to dementia is a major pursuit in current research. It is broadly accepted that cognition declines with a continuum between MCI and dementia. As such, cohorts of MCI patients are usually heterogeneous, containing patients at different stages of the neurodegenerative process. This hampers the prognostic task. Nevertheless, when learning prognostic models, most studies use the entire cohort of MCI patients regardless of their disease stages. In this paper, we propose a Time Windows approach to predict conversion to dementia, learning with patients stratified using time windows, thus fine-tuning the prognosis regarding the time to conversion. In the proposed Time Windows approach, we grouped patients based on the clinical information of whether they converted (converter MCI) or remained MCI (stable MCI) within a specific time window. We tested time windows of 2, 3, 4 and 5 years. We developed a prognostic model for each time window using clinical and neuropsychological data and compared this approach with the commonly used in the literature, where all patients are used to learn the models, named as First Last approach. This enables to move from the traditional question “Will a MCI patient convert to dementia somewhere in the future” to the question “Will a MCI patient convert to dementia in a specific time window”. The proposed Time Windows approach outperformed the First Last approach. The results showed that we can predict conversion to dementia as early as 5 years before the event with an AUC of 0.88 in the cross-validation set and 0.76 in an independent validation set. Prognostic models using time windows have higher performance when predicting progression from MCI to dementia, when compared to the prognostic approach commonly used in the literature. Furthermore, the proposed Time Windows approach is more relevant from a clinical point of view, predicting conversion within a temporal interval rather than sometime in the future and allowing clinicians to timely adjust treatments and clinical appointments.

Unifying thread-level speculation and transactional memory

Abstract: The motivation of this work is to ask whether Transactional Memory (TM) and Thread-Level Speculation (TLS), two prominent concurrency paradigms usually considered separately, can be combined into a hybrid approach that extracts untapped parallelism and speed-up from common programs. We show that the answer is positive by describing an algorithm, called TLSTM, that leverages an existing TM with TLS capabilities. We also show that our approach is able to achieve up to a 48% increase in throughput over the base TM, on read dominated workloads of long transactions in a multi-threaded application, among other results.

MISNIS: An intelligent platform for twitter topic mining

Abstract: Twitter has become a major tool for spreading news, for dissemination of positions and ideas, and for the commenting and analysis of current world events. However, with more than 500 million tweets flowing per day, it is necessary to find efficient ways of collecting, storing, managing, mining and visualizing all this information. This is especially relevant if one considers that Twitter has no ways of indexing tweet contents, and that the only available categorization “mechanism” is the #hashtag, which is totally dependent of a user's will to use it. This paper presents an intelligent platform and framework, named MISNIS - Intelligent Mining of Public Social Networks’ Influence in Society - that facilitates these issues and allows a non-technical user to easily mine a given topic from a very large tweet's corpus and obtain relevant contents and indicators such as user influence or sentiment analysis. When compared to other existent similar platforms, MISNIS is an expert system that includes specifically developed intelligent techniques that: (1) Circumvent the Twitter API restrictions that limit access to 1% of all flowing tweets. The platform has been able to collect more than 80% of all flowing portuguese language tweets in Portugal when online; (2) Intelligently retrieve most tweets related to a given topic even when the tweets do not contain the topic #hashtag or user indicated keywords. A 40% increase in the number of retrieved relevant tweets has been reported in real world case studies. The platform is currently focused on Portuguese language tweets posted in Portugal. However, most developed technologies are language independent (e.g. intelligent retrieval, sentiment analysis, etc.), and technically MISNIS can be easily expanded to cover other languages and locations.