Research

PRHLT

The Pattern Recognition and Human Language Technology PRHLT research center is composed by researchers from the Universidad PolitÚcnica de Valencia (UPV) in the areas of Multimodal Interaction, Pattern Recognition, Image Processing (Image Analysis, Computer Vision, Handwritten Text Recognition, Document Analysis) and Language Processing (Speech Recognition and Understanding, Machine Translation, Information Retrieval). The PRHLT center is an active research entity with important ongoing research projects, technology transfer activities, and research publications.


My research topics

  • Machine Learning
  • Statistical Pattern Recognition
  • Computer Vision


  • Neural Networks, Deep Learning
  • Social Analysis, Community Detection
  • Nearest Neighbors, Metric Learning
  • Image Retrieval, Relevance Feedback
  • Biometrics, Facial Analysis

My Online Learning lecture is one of the top voted Online Learning lectures in VideoLectues:


Online Learning

Roberto Paredes
4ávideos


My Google Scholar profile: Roberto Paredes



Some of my research results:

Our Feature representation for social circles detection using MAC paper has been accepted in Neural Computing and Applications


MAC approach

Social circles detection is a special case of community detection in social network that is currently attracting a growing interest in the research community. In this paper, we propose an empirical evaluation of the multi-assignment clustering method using different feature representation models. We define different vectorial representations from both structural egonet information and user profile features. We study and compare the performance on two available labelled Facebook datasets and compare our results with several different baselines. In addition, we provide some insights of the evaluation metrics most commonly used in the literature.


Our Local Deep Neural Networks paper has been accepted in Pattern Recognition Letters


LDNN approach

Deep Learning methods are able to automatically discover better representations of the data to improve the performance of the classifiers. However, in computer vision tasks, such us the gender recognition problem, sometimes it is difficult to directly learn from the entire image. In this work we propose a new model called Local Deep Neural Network (Local-DNN), which is based on two key concepts: local features and deep architectures. The model learns from small overlapping regions in the visual field using discriminative feed-forward networks with several layers. We evaluate our approach on two well-known gender benchmarks, showing that our Local-DNN outperforms other deep learning methods also evaluated and obtains state-of-the-art results in both benchmarks.


Our paper on selective regularizarion for Restricted Boltzmann Machines appears in Neurocomputing


LDNN approach

In this work we propose to deal with two important issues regarding to the RBMs learning capabilities. First, the topology of the input space, and second, the sparseness of the RBM obtained. One problem of RBMs is that they do not take advantage of the topology of the input space. In order to alleviate this lack, we propose to use a surrogate of the mutual information of the input representation space to build a set of binary masks. This approach is general and not only applicable to images, thus it can be extended to other layers in the standard layer-by-layer unsupervised learning. On the other hand, we propose a selective application of two different regularization terms, L1 and L2, in order to ensure the sparseness of the representation and the generalization capabilities. Additionally, another interesting capability of our approach is the adaptation of the topology of the network during the learning phase by means of selecting the best set of binary masks that fit the current weights configuration. The performance of these new ideas is assessed with a set of experiments on different well-known corpus.


Our paper on Feature Representation for Social Circles accepted in IbPria 2015


LDNN approach

Social circles detection is a special case of community detection in social network that is currently attracting a growing interest in the research community. We propose in this paper an empirical evaluation of the multi-assignment clustering method using different feature representation models. We define different vectorial representations from both structural egonet information and user profile features. We study and compare the performance on the available labelled Facebook data from the Kaggle competition on learning social circles in networks. We compare our results with several different baselines.