A multimodal genre recognition prototype

In this paper, a multimodal and interactive prototype to perform music genre classification is presented. The system is oriented to multi-part files in symbolic format but it can be adapted using a transcription system to transform audio content in music scores. This prototype uses different sources of information to give a possible answer to the user. It has been developed to allow a human expert to interact with the system to improve its results. In its current implementation, it offers a limited range of interaction and multimodality. Further development aimed at full interactivity and multimodal interactions is discussed.

Recognition of online handwritten music symbols

Paper submitted to MML 2013, 6th International Workshop on Machine Learning and Music, Prague, September 23, 2013.

Evaluation of an integrated system for classification, assessment and comparison of services for long-term care in Europe: the eDESDE-LTC study

Background: The harmonization of European health systems brings with it a need for tools to allow the standardized collection of information about medical care. A common coding system and standards for the description of services are needed to allow local data to be incorporated into evidence-informed policy, and to permit equity and mobility to be assessed. The aim of this project has been to design such a classification and a related tool for the coding of services for Long Term Care (DESDE-LTC), based on the European Service Mapping Schedule (ESMS). Methods: The development of DESDE-LTC followed an iterative process using nominal groups in 6 European countries. 54 researchers and stakeholders in health and social services contributed to this process. In order to classify services, we use the minimal organization unit or “Basic Stable Input of Care” (BSIC), coded by its principal function or “Main Type of Care” (MTC). The evaluation of the tool included an analysis of feasibility, consistency, ontology, inter-rater reliability, Boolean Factor Analysis, and a preliminary impact analysis (screening, scoping and appraisal). Results: DESDE-LTC includes an alpha-numerical coding system, a glossary and an assessment instrument for mapping and counting LTC. It shows high feasibility, consistency, inter-rater reliability and face, content and construct validity. DESDE-LTC is ontologically consistent. It is regarded by experts as useful and relevant for evidence-informed decision making. Conclusion: DESDE-LTC contributes to establishing a common terminology, taxonomy and coding of LTC services in a European context, and a standard procedure for data collection and international comparison.

Improving 3D Keypoint Detection from Noisy Data Using Growing Neural Gas

3D sensors provides valuable information for mobile robotic tasks like scene classification or object recognition, but these sensors often produce noisy data that makes impossible applying classical keypoint detection and feature extraction techniques. Therefore, noise removal and downsampling have become essential steps in 3D data processing. In this work, we propose the use of a 3D filtering and down-sampling technique based on a Growing Neural Gas (GNG) network. GNG method is able to deal with outliers presents in the input data. These features allows to represent 3D spaces, obtaining an induced Delaunay Triangulation of the input space. Experiments show how the state-of-the-art keypoint detectors improve their performance using GNG output representation as input data. Descriptors extracted on improved keypoints perform better matching in robotics applications as 3D scene registration.

Improving kNN multi-label classification in Prototype Selection scenarios using class proposals

Prototype Selection (PS) algorithms allow a faster Nearest Neighbor classification by keeping only the most profitable prototypes of the training set. In turn, these schemes typically lower the performance accuracy. In this work a new strategy for multi-label classifications tasks is proposed to solve this accuracy drop without the need of using all the training set. For that, given a new instance, the PS algorithm is used as a fast recommender system which retrieves the most likely classes. Then, the actual classification is performed only considering the prototypes from the initial training set belonging to the suggested classes. Results show that this strategy provides a large set of trade-off solutions which fills the gap between PS-based classification efficiency and conventional kNN accuracy. Furthermore, this scheme is not only able to, at best, reach the performance of conventional kNN with barely a third of distances computed, but it does also outperform the latter in noisy scenarios, proving to be a much more robust approach.

Application of multi-sensor advanced DInSAR analysis to severe land subsidence recognition: Alto Guadalentín Basin (Spain)

Multi-sensor advanced DInSAR analyses have been performed and compared with two GPS station measurements, in order to evaluate the land subsidence evolution in a 20-year period, in the Alto Guadalentín Basin where the highest rate of man-induced subsidence (> 10 cm yr−1) of Europe had been detected. The control mechanisms have been examined comparing the advanced DInSAR data with conditioning and triggering factors (i.e. isobaths of Plio-Quaternary deposits, soft soil thickness and piezometric level).

Human Behaviour Recognition based on Trajectory Analysis using Neural Networks

Automated human behaviour analysis has been, and still remains, a challenging problem. It has been dealt from different points of views: from primitive actions to human interaction recognition. This paper is focused on trajectory analysis which allows a simple high level understanding of complex human behaviour. It is proposed a novel representation method of trajectory data, called Activity Description Vector (ADV) based on the number of occurrences of a person is in a specific point of the scenario and the local movements that perform in it. The ADV is calculated for each cell of the scenario in which it is spatially sampled obtaining a cue for different clustering methods. The ADV representation has been tested as the input of several classic classifiers and compared to other approaches using CAVIAR dataset sequences obtaining great accuracy in the recognition of the behaviour of people in a Shopping Centre.

On the suitability of Prototype Selection methods for kNN classification with distributed data

In the current Information Age, data production and processing demands are ever increasing. This has motivated the appearance of large-scale distributed information. This phenomenon also applies to Pattern Recognition so that classic and common algorithms, such as the k-Nearest Neighbour, are unable to be used. To improve the efficiency of this classifier, Prototype Selection (PS) strategies can be used. Nevertheless, current PS algorithms were not designed to deal with distributed data, and their performance is therefore unknown under these conditions. This work is devoted to carrying out an experimental study on a simulated framework in which PS strategies can be compared under classical conditions as well as those expected in distributed scenarios. Our results report a general behaviour that is degraded as conditions approach to more realistic scenarios. However, our experiments also show that some methods are able to achieve a fairly similar performance to that of the non-distributed scenario. Thus, although there is a clear need for developing specific PS methodologies and algorithms for tackling these situations, those that reported a higher robustness against such conditions may be good candidates from which to start.

#DressCodePM & Re-Framing The Niqab: News Sources, Hashtag Activism, and Media Representation

This thesis originates from my interest in exploring how minorities are using social media to talk back to mainstream media. This study examines whether hashtags that trend on Twitter may impact how news stories related to minorities are covered in Canadian media. The Canadian Prime Minister Stephen Harper stated the niqab was “rooted in a culture that is anti-women” on 10 March 2015. The next day #DressCodePM trended in response to the PM’s niqab remarks. Using network gatekeeping theory, this study examines the types of sources quoted in the media stories published on 10 and 11 March 2015. The study’s goal is to explore whether using tweet quotes leads to the representation of a more diverse range of news sources. The study compares the types of sources quoted in stories that covered Harper’s comments without mentioning #DressCodePM versus stories that mention #DressCodePM. This study also uses Tuen A. van Dijk’s methodology of asking “who is speaking, how often and how prominently?” in order to examine whose voices have been privileged and whose voices have been marginalized in covering the niqab in Canadian media from the 1970s and until the days following the PM’s remarks. Network gatekeeping theory is applied in this study to assess whether the gated gained more power after #DressCodePM trended. The case study’s findings indicates that Caucasian male politicians were predominantly used as news sources in covering stories related to the niqab for the past 38 years in the Globe and Mail. The sourcing pattern of favouring politicians continued in Canadian print and online media on 10 March 2015 following Harper’s niqab comments. However, ordinary Canadian women, including Muslim women, were used more often than politicians as news sources in the stories about #DressCodePM that were published on 11 March 2015. The gated media users were able to gain power and attract Canadian Media’s attention by widely spreading #DressCodePM. This study draws attention to the lack of diversity of sources used in Canadian political news stories, yet this study also shows it is possible for the gated media users to amplify their voices through hashtag activism.

Culture & ethnology,

First published 1917. "Reprinted March, 1929 by permission of Horace Liveright."

Beauty illustrated by an analysis and classification of beauty in woman : with a critical view of the hypotheses of Hume, Hogarth, Burke, Knight, Alison, etc., and the hypotheses of beauty in sculpture and painting, by Leonardo da Vinci, Winckelmann, Mengs, Bossi, etc. /

The plates include one frontispiece, 22 plates numbered I-XXII and one addtional plate numbered XVIb.

Albanian Sworn Virgins: A Visual Representation of their Transformation

This piece of art is a flipbook, analogous to the ones children play with as they make cartoon balls bounce with the quick flipping of pages between their thumb and index finger. However, instead of a playful scene, this flipbook is a commentary on Albanian Sworn Virgins. These are women from Northern Albania who, in their youth, swear to celibacy in order to gain the societal power that is exclusive to men in their culture. This flipbook demonstrates this cultural male-to-female shift and comments on its inability to ever be fully realized. This commentary is inspired by the words of Albanian Sworn Virgins in Elvira Dones’ documentary, Sworn Virgins, who feel betrayed by their biological need to menstruate and who view their reproductive system as a permanent obstacle in completing their societal shift. Just as a child’s flipbook tells a story, this flipbook illustrates the Albanian Sworn Virgins’ forever-unfinished transformation.

New Techniques in Deep Representation Learning

Thesis (Ph.D.)--University of Washington, 2016-06

Marine mammal species detection and classification

Thesis (Ph.D.)--University of Washington, 2016-04

A longitudinal investigation of self-other discrimination and the emergence of mirror self-recognition

The aim in this study was to investigate the association between infants' developing interest in their self-image and the onset of mirror self-recognition (MSR). A longitudinal study was conducted with 98 infants who were seen at intervals of 3 months from 9-24 months of age. At each session, the infants were administered a preferential-looking test whereby they were presented with a video image of themselves alongside a video image of a same-aged peer in two conditions, unmarked and marked. From the 12-month session onwards, the infants were also administered a version of the standard mark test of MSR. The infants showed a significant preference for looking at images of themselves in both conditions coincident with the onset of MSR. This result indicates that developing an interest in the self-image is an important component in the development of MSR. (C) 2003 Elsevier Science Inc. All rights reserved.