Claudio A. Perez

Normalized Mutual Information Feature Selection

A filter method of feature selection based on mutual information, called normalized mutual information feature selection (NMIFS), is presented. NMIFS is an enhancement over Battitis MIFS, MIFS-U, and mRMR methods. The average normalized mutual information is proposed as a measure of redundancy among features. NMIFS outperformed MIFS, MIFS-U, and mRMR on several artificial and benchmark data sets without requiring a user-defined parameter. In addition, NMIFS is combined with a genetic algorithm to form a hybrid filter/wrapper method called GAMIFS. This includes an initialization procedure and a mutation operator based on NMIFS to speed up the convergence of the genetic algorithm. GAMIFS overcomes the limitations of incremental search algorithms that are unable to find dependencies between groups of features.

Dendritic cell immunizations alone or combined with low doses of interleukin-2 induce specific immune responses in melanoma patients.

Dendritic cell (DC)‐based therapy has proved to be effective in patients with a variety of malignancies. However, an optimal immunization protocol using DCs and the best means for delivering antigens has not yet been described. In this study, 20 patients with malignant melanoma in stages III or IV were vaccinated with autologous DCs pulsed with a melanoma cell lysate, alone (n = 13) or in combination with low doses of subcutaneous (s.c.) interleukin (IL)‐2 injections (n = 7), to assess toxicity, immunological and clinical responses. Monocyte‐derived DCs were morphological, phenotypic and functionally characterized in vitro. Peripheral blood mononuclear cells (PBMC), harvested from patients either prior to and after the treatment, were analysed using enzyme‐linked immunosorbent spot (ELISPOT). After vaccination, 50% of the patients tested (seven of 13) from the first group and (three of seven) from the second, showed an increase in interferon (IFN)‐γ production in response to allogeneic melanoma cell lines but not to controls. Four of five tested human leucocyte antigen (HLA)‐A2+ patients with anti‐melanoma activity also showed specific T cell responses against peptides derived from melanoma‐associated antigens. Delayed type IV hypersensitivity reaction (DTH) against melanoma cell lysate was observed in six of 13 patients from the group treated with DC vaccines only and four of seven from the group treated with the combination of DCs and IL‐2. Significant correlations were found between DTH‐positive responses against tumour lysate and both disease stability and post‐vaccination survival on the stage IV patients. There were no toxicities associated with the vaccines or evidence of autoimmunity including vitiligo. Furthermore, no significant enhancement was observed as a result of combining DC vaccination with IL‐2. Our data suggest that autologous DCs pulsed with tumour lysate may provide a standardized and widely applicable source of melanoma specific antigens for clinical use. It is safe and causes no significant side effects and has been demonstrated to be partially efficient at triggering effective anti‐melanoma immunity.

Subscription fraud prevention in telecommunications using fuzzy rules and neural networks

A system to prevent subscription fraud in fixed telecommunications with high impact on long-distance carriers is proposed. The system consists of a classification module and a prediction module. The classification module classifies subscribers according to their previous historical behavior into four different categories: subscription fraudulent, otherwise fraudulent, insolvent and normal. The prediction module allows us to identify potential fraudulent customers at the time of subscription. The classification module was implemented using fuzzy rules. It was applied to a database containing information of over 10,000 real subscribers of a major telecom company in Chile. In this database a subscription fraud prevalence of 2.2% was found. The prediction module was implemented as a multilayer perceptron neural network. It was able to identify 56.2% of the true fraudsters, screening only 3.5% of all the subscribers in the test set. This study shows the feasibility of significantly preventing subscription fraud in telecommunications by analyzing the application information and the customer antecedents at the time of application.

Gender Classification Based on Fusion of Different Spatial Scale Features Selected by Mutual Information From Histogram of LBP, Intensity, and Shape

In this paper, we report our extension of the use of feature selection based on mutual information and feature fusion to improve gender classification of face images. We compare the results of fusing three groups of features, three spatial scales, and four different mutual information measures to select features. We also showed improved results by fusion of LBP features with different radii and spatial scales, and the selection of features using mutual information. As measures of mutual information we use minimum redundancy and maximal relevance (mRMR), normalized mutual information feature selection (NMIFS), conditional mutual information feature selection (CMIFS), and conditional mutual information maximization (CMIM). We tested the results on four databases: FERET and UND, under controlled conditions, the LFW database under unconstrained scenarios, and AR for occlusions. It is shown that selection of features together with fusion of LBP features significantly improved gender classification accuracy compared to previously published results. We also show a significant reduction in processing time because of the feature selection, which makes real-time applications of gender classification feasible.

Tumour necrosis factor-α (tnf-α) levels and influence of -308 TNF-α promoter polymorphism on the responsiveness to infliximab in patients with rheumatoid arthritis

Objective: To investigate the influence of −308 tumour necrosis factor‐α (TNF‐α) promoter polymorphism and circulating TNF‐α levels in the clinical response to the infliximab treatment in patients with rheumatoid arthritis (RA). Methods: One hundred and thirty‐two RA patients were genotyped for TNF‐α promoter by polymerase‐chain reaction restriction fragment‐length polymorphism (PCR‐RFLP) analysis. Ten patients with the −308 TNF‐α gene promoter genotype G/A, and 10 with the G/G genotype were selected and received 3 mg/kg of infliximab at Weeks 0, 2, 6, and 14. Results: Both groups showed a significant improvement with treatment in all variables studied. Total mean TNF‐α levels increased significantly with respect to basal levels in most of patients after treatment [probability (p)=0.04]. Only patients from G/A showed a statistically significant correlation between ACR 50 and the increase of TNF‐α levels (p<0.03). Conclusion: A relationship was detected between ACR criteria of improvement and increased circulating TNF‐α levels in RA patients subjected to anti‐TNF‐α therapy.

Methodological improvement on local Gabor face recognition based on feature selection and enhanced Borda count

Face recognition has a wide range of possible applications in surveillance, human computer interfaces and marketing and advertising goods for selected customers according to age and gender. Because of the high classification rate and reduced computational time, one of the best methods for face recognition is based on Gabor jet feature extraction and Borda count classification. In this paper, we propose methodological improvements to increase face recognition rate by selection of Gabor jets using entropy and genetic algorithms. This selection of jets additionally allows faster processing for real-time face recognition. We also propose improvements in the Borda count classification through a weighted Borda count and a threshold to eliminate low score jets from the voting process to increase the face recognition rate. Combinations of Gabor jet selection and Borda count improvements are also proposed. We compare our results with those published in the literature to date and find significant improvements. Our best results on the FERET database are 99.8%, 99.5%, 89.2% and 86.8% recognition rates on the subsets Fb, Fc, Dup1 and Dup2, respectively. Compared to the best results published in the literature, the total number of recognition errors decreased from 163 to 112 (31%). We also tested the proposed method under illumination changes, occlusions with sunglasses and scarves and for small pose variations. Results on two different face databases (AR and Extended Yale B) with significant illumination changes showed over 90% recognition rate. The combination EJS-BTH-BIP reached 98% and 99% recognition rate in images with sunglasses and scarves from the AR database, respectively. The proposed method reached 93.5% recognition on faces with small pose variation of 25? rotation and 98.5% with 15% rotation in the FERET database.

Face and eye tracking algorithm based on digital image processing

A non-invasive interface to track eye position using digital image processing techniques is under development. Information about head and eye position is obtained from digital images. The objective is to develop an interface to detect eye position based only on digital image processing algorithms, free of electrodes or other electronic devices. We propose a method for eye tracking built into five stages. These include: coarse and fine face detection, finding the eye region of maximum probability, map of the pupil/iris location and pupil/iris detection. Using frontal face images obtained from a database, the probability maps for the eye region were built. Only gray levels are considered for this computation (8 bits). The algorithms for face and eye detection were assessed on 102 images from the Purdue database and on 897 images from a video sequence. The face detection algorithm reached a 99% and 100% correct detection rate on the databases respectively. On the same databases the pupil/iris detection algorithm reached 85.3% and 98.4% of correct detection respectively.

Coumarin-based fluorescent probes for dual recognition of copper(II) and iron(III) ions and their application in bio-imaging.

Two new coumarin-based “turn-off” fluorescent probes, (E)-3-((3,4-dihydroxybenzylidene)amino)-7-hydroxy-2H-chromen-2-one (BS1) and (E)-3-((2,4-dihydroxybenzylidene)amino)-7-hydroxy-2H-chromen-2-one (BS2), were synthesized and their detection of copper(II) and iron(III) ions was studied. Results show that both compounds are highly selective for Cu2+ and Fe3+ ions over other metal ions. However, BS2 is detected directly, while detection of BS1 involves a hydrolysis reaction to regenerate 3-amino-7-hydroxycoumarin (3) and 3,4-dihydroxybenzaldehyde, of which 3 is able to react with copper(II) or iron(III) ions. The interaction between the tested compounds and copper or iron ions is associated with a large fluorescence decrease, showing detection limits of ca. 10−5 M. Preliminary studies employing epifluorescence microscopy demonstrate that Cu2+ and Fe3+ ions can be imaged in human neuroblastoma SH-SY5Y cells treated with the tested probes.

Extracting Fuzzy Rules From Polysomnographic Recordings for Infant Sleep Classification

A neuro-fuzzy classifier (NFC) of sleep-wake states and stages has been developed for healthy infants of ages 6 mo and onward. The NFC takes five input patterns previously identified on 20-s epochs from polysomnographic recordings and assigns them to one out of five possible classes: Wakefulness, REM-Sleep, Non-REM Sleep Stage 1, Stage 2, and Stage 3-4. The definite criterion for a sleep state or stage to be established is duration of at least 1 min. The data set consisted of a total of 14 continuous recordings of naturally occurring naps (average duration: 143plusmn39 min), corresponding to a total of 6021 epochs. They were divided in a training, a validation and a test set with 7, 2, and 5 recordings, respectively. During supervised training, the system determined the fuzzy concepts associated to the inputs and the rules required for performing the classification, extracting knowledge from the training set, and pruning nonrelevant rules. Results on an independent test set achieved 83.9plusmn0.4% of expert agreement. The fuzzy rules obtained from the training examples without a priori information showed a high level of coincidence with the crisp rules stated by the experts, which are based on internationally accepted criteria. These results show that the NFC can be a valuable tool for implementing an automated sleep-wake classification system

Genetic design of biologically inspired receptive fields for neural pattern recognition

This paper proposes a new method for the design, through simulated evolution, of biologically inspired receptive fields in feedforward neural networks (NNs). The method is intended to enhance pattern recognition performance by creating new neural architectures specifically tuned for a particular pattern recognition problem. It proposes a combined neural architecture composed of two networks in cascade: a feature extraction network (FEN) followed by a neural classifier. The FEN is composed of several layers with receptive fields constructed by additive superposition of excitatory and inhibitory fields. A genetic algorithm (GA) is used to select receptive field parameters to improve classification performance. The parameters are receptive field size, orientation, and bias as well as the number of different receptive fields in each layer. Based on a random initial population where each individual represents a different neural architecture, the GA creates new enhanced individuals. The method is applied to handwritten digit classification and face recognition. In both problems, results show strong dependency between NN classification performance and receptive field architecture. GA selected parameters of the receptive fields produced improvements in the classification performance on the test set up to 90.8% for the problem of handwritten digit classification and up to 84.2% for the face recognition problem. On the same test sets, results were compared advantageously to standard feedforward multilayer perceptron (MLP) NNs where receptive fields are not explicitly defined. The MLP reached a maximum classification performance of 84.9% and 77.5% in both problems, respectively.