Search results

8,557 records were found.

We use Kernel Canonical Correlation Analysis (KCCA) to infer brain activity in functional MRI by learning a semantic representation of fMRI brain scans and their associated activity signal. The semantic space provides a common representation and enables a comparison between the fMRI and the activity signal. We compare the approach against Canonical Correlation Analysis (CCA) and the more commonly used Ordinary Correlation Analysis (OCA) by localising “activity” on a simulated null data set. We also compare performance of the methods on the localisation of brain regions which control finger movement and regions that are involved in mental calculation. Finally we present an approach to reconstruct an activity signal from an “unknown” testing-set fMRI scans. This is used to validate the learnt semantics as non-trivial.
Functional dependencies, a notion originated in Relational Database Theory, are known to admit interesting characterizations in terms of Formal Concept Analysis. In database terms, two successive, natural extensions of the notion of functional dependency are the so-called degenerate multivalued dependencies, and multivalued dependencies proper. We propose here a new Galois connection, based on any given relation, which gives rise to a formal concept lattice corresponding precisely to the degenerate multivalued dependencies that hold in the relation given. The general form of the construction departs significantly from the most usual way of handling functional dependencies. Then, we extend our approach so as to extract Armstrong relations for the degenerate multivalued dependencies from the concept lattice obtained; the proof of the c...
We consider $n$ anonymous selfish users that route their communication through $m$ parallel links. The users are allowed to reroute, concurrently, from overloaded links to underloaded links. The different rerouting decisions are concurrent, randomized and independent. The rerouting process terminates when the system reaches a Nash equilibrium, in which no user can improve its state. We study the convergence rate of several migration policies. The first is a very natural policy, which balances the expected load on the links, for the case that all users are identical and apply it, we show that the rerouting terminates in expected $O(\log \log n + \log^{1.5} m)$ stages. The second policy that we consider is a Nash rerouting policy, which is a myopic strategic policy, in which {\em every} rerouting stage is a Nash equilibrium. For the Nas...
We present a concise tutorial on statistical learning, the theoretical ground on which the learning from examples paradigm is based. We also discuss the problem of face detection as a case study illustrating the solutions proposed in this framework. Finally, we describe some new results we obtained by means of an object detection method based on statistical hypothesis tests which makes use of positive examples only.
We present a new approximation scheme for support vector decision functions in object detection. In the present approach we are building on an existing algorithm where the set of support vectors is replaced by a smaller so-called reduced set of synthetic points. Instead of finding the reduced set via unconstrained optimization, we impose a structural constraint on the synthetic vectors such that the resulting approximation can be evaluated via separable filters. Applications that require scanning an entire image can benefit from this representation: when using separable filters, the average computational complexity for evaluating a reduced set vector on a test patch of size (h x w) drops from O(hw) to O(h+w). We show experimental results on handwritten digits and face detection.
Boosting is a general method for training an ensemble of classifiers with a view to improving performance relative to that of a single classifier. While the original AdaBoost algorithm has been defined for classification tasks, the current work examines its applicability to sequence learning problems. In particular, different methods for training HMMs on sequences and for combining their output are investigated in the context of automatic speech recognition.
Ce travail concerne le développement de méthodes de classification discriminantes pour des données séquentielles. Quelques techniques ont été proposées pour étendre aux séquences les méthodes discriminantes, comme les machines à vecteurs supports, par nature plus adaptées aux données en dimension fixe. Elles permettent de classifier des séquences complètes mais pas de réaliser la segmentation, qui consiste à reconnaître la séquence d’unités, phonèmes ou lettres par exemple, correspondant à un signal. En utilisant une correspondance donnée / modèle nous transformons le problème de l’apprentissage des modèles à partir de données par un problème de sélection de modèles, qui peut être attaqué via des méthodes du type machines à vecteurs supports. Nous proposons et évaluons divers noyaux pour cela et fournissons des résultats expérimentaux ...
Querying heterogeneous XML document collections is an open problem. This will require building some sort of correspondence between the DTD of the different sources. We consider here the problem of matching the structure of XML documents from different sources. We introduce for that a stochastic structured document model and describe preliminary experiments performed on the INEX collection.
A selective sampling algorithm is a learning algorithm for classification that, based on the past observed data, decides whether to ask the label of each new instance to be classified. In this paper, we introduce a general technique for turning linear-threshold classification algorithms from the general additive family into randomized selective sampling algorithms. For the most popular algorithms in this family we derive mistake bounds that hold for individual sequences of examples. These bounds show that our semi-supervised algorithms can achieve, on average, the same accuracy as that of their fully supervised counterparts, but using fewer labels. Our theoretical results are corroborated by a number of experiments on real-world textual data. The outcome of these experiments is essentially predicted by our theoretical results: Our sel...
This work concerns note-taking applications; it deals with poorly structured on-line handwritten documents segmentation such as pages of handwritten notes. We extend an existing system based on Probabilistic Feature Grammars. The probabilistic nature of this system allows considering lots of segmentation hypothesis, which is an advantage for poorly structured documents processing, but it goes with important algorithmic complexity. Our improvements concern the handling of this complexity using genetic algorithms, the definition of performance measurements that are adapted to the segmentation of on-line documents, and the evaluation of this segmentation approach on a collection of documents of various quality.
Want to know more?If you want to know more about this cutting edge product, or schedule a demonstration on your own organisation, please feel free to contact us or read the available documentation at http://www.keep.pt/produtos/retrievo/?lang=en