Enhancing Multimodal Silent Speech Interfaces With Feature Selection

In research on Silent Speech Interfaces (SSI), different sources of information (modalities) have been combined, aiming at obtaining better performance than the individual modalities. However, when combining these modalities, the dimensionality of the feature space rapidly increases, yielding the well-known "curse of dimensionality". As a consequence, in order to extract useful information from this data, one has to resort to feature selection (FS) techniques to lower the dimensionality of the learning space. In this paper, we assess the impact of FS techniques for silent speech data, in a dataset with 4 non-invasive and promising modalities, namely: video, depth, ultrasonic Doppler sensing, and surface electromyography. We consider two supervised (mutual information and Fisher's ratio) and two unsupervised (meanmedian and arithmetic mean geometric mean) FS filters. The evaluation was made by assessing the classification accuracy (word recognition error) of three well-known classifiers (knearest neighbors, support vector machines, and dynamic time warping). The key results of this study show that both unsupervised and supervised FS techniques improve on the classification accuracy on both individual and combined modalities. For instance, on the video component, we attain relative performance gains of 36.2% in error rates. FS is also useful as pre-processing for feature fusion. Copyright © 2014 ISCA.

Model-based Segmentation and Image Fusion of 3D Computed Tomography and 3D Ultrasound of the Eye for Radiotherapy Planning

For radiotherapy treatment planning of retinoblastoma inchildhood, Computed Tomography (CT) represents thestandard method for tumor volume delineation, despitesome inherent limitations. CT scan is very useful inproviding information on physical density for dosecalculation and morphological volumetric information butpresents a low sensitivity in assessing the tumorviability. On the other hand, 3D ultrasound (US) allows ahigh accurate definition of the tumor volume thanks toits high spatial resolution but it is not currentlyintegrated in the treatment planning but used only fordiagnosis and follow-up. Our ultimate goal is anautomatic segmentation of gross tumor volume (GTV) in the3D US, the segmentation of the organs at risk (OAR) inthe CT and the registration of both. In this paper, wepresent some preliminary results in this direction. Wepresent 3D active contour-based segmentation of the eyeball and the lens in CT images; the presented approachincorporates the prior knowledge of the anatomy by usinga 3D geometrical eye model. The automated segmentationresults are validated by comparing with manualsegmentations. Then, for the fusion of 3D CT and USimages, we present two approaches: (i) landmark-basedtransformation, and (ii) object-based transformation thatmakes use of eye ball contour information on CT and USimages.

Maximum likelihood Linear Programming Data Fusion for Speaker Recognition

Biometric system performance can be improved by means of data fusion. Several kinds of information can be fused in order to obtain a more accurate classification (identification or verification) of an input sample. In this paper we present a method for computing the weights in a weighted sum fusion for score combinations, by means of a likelihood model. The maximum likelihood estimation is set as a linear programming problem. The scores are derived from a GMM classifier working on a different feature extractor. Our experimental results assesed the robustness of the system in front a changes on time (different sessions) and robustness in front a change of microphone. The improvements obtained were significantly better (error bars of two standard deviations) than a uniform weighted sum or a uniform weighted product or the best single classifier. The proposed method scales computationaly with the number of scores to be fussioned as the simplex method for linear programming.

Social network extraction and analysis based on multimodal dyadic interaction

Social interactions are a very important component in people"s lives. Social network analysis has become a common technique used to model and quantify the properties of social interactions. In this paper, we propose an integrated framework to explore the characteristics of a social network extracted from multimodal dyadic interactions. For our study, we used a set of videos belonging to New York Times" Blogging Heads opinion blog. The Social Network is represented as an oriented graph, whose directed links are determined by the Influence Model. The links" weights are a measure of the"influence" a person has over the other. The states of the Influence Model encode automatically extracted audio/visual features from our videos using state-of-the art algorithms. Our results are reported in terms of accuracy of audio/visual data fusion for speaker segmentation and centrality measures used to characterize the extracted social network.

Multi-level Input Fusion in multi-modal systems

Context-aware multimodal interactive systems aim to adapt to the needs and behavioural patterns of users and offer a way forward for enhancing the efficacy and quality of experience (QoE) in human-computer interaction. The various modalities that constribute to such systems each provide a specific uni-modal response that is integratively presented as a multi-modal interface capable of interpretation of multi-modal user input and appropriately responding to it through dynamically adapted multi-modal interactive flow management , This paper presents an initial background study in the context of the first phase of a PhD research programme in the area of optimisation of data fusion techniques to serve multimodal interactivite systems, their applications and requirements.

Towards anomaly detection for increased security in multibiometric systems: spoofing-resistant 1-median fusion eliminating outliers

Multibiometrics aims at improving biometric security in presence of spoofing attempts, but exposes a larger availability of points of attack. Standard fusion rules have been shown to be highly sensitive to spoofing attempts – even in case of a single fake instance only. This paper presents a novel spoofing-resistant fusion scheme proposing the detection and elimination of anomalous fusion input in an ensemble of evidence with liveness information. This approach aims at making multibiometric systems more resistant to presentation attacks by modeling the typical behaviour of human surveillance operators detecting anomalies as employed in many decision support systems. It is shown to improve security, while retaining the high accuracy level of standard fusion approaches on the latest Fingerprint Liveness Detection Competition (LivDet) 2013 dataset.

Fusion of fingerprint recognition methods for robust human identification

Biometrics is one of the biggest tendencies in human identification. The fingerprint is the most widely used biometric. However considering the automatic fingerprint recognition a completely solved problem is a common mistake. The most popular and extensively used methods, the minutiae-based, do not perform well on poor-quality images and when just a small area of overlap between the template and the query images exists. The use of multibiometrics is considered one of the keys to overcome the weakness and improve the accuracy of biometrics systems. This paper presents the fusion of a minutiae-based and a ridge-based fingerprint recognition method at rank, decision and score level. The fusion techniques implemented leaded to a reduction of the Equal Error Rate by 31.78% (from 4.09% to 2.79%) and a decreasing of 6 positions in the rank to reach a Correct Retrieval (from rank 8 to 2) when assessed in the FVC2002-DB1A database. © 2008 IEEE.

A rank aggregation framework for video multimodal geocoding

This paper proposes a rank aggregation framework for video multimodal geocoding. Textual and visual descriptions associated with videos are used to define ranked lists. These ranked lists are later combined, and the resulting ranked list is used to define appropriate locations for videos. An architecture that implements the proposed framework is designed. In this architecture, there are specific modules for each modality (e.g, textual and visual) that can be developed and evolved independently. Another component is a data fusion module responsible for combining seamlessly the ranked lists defined for each modality. We have validated the proposed framework in the context of the MediaEval 2012 Placing Task, whose objective is to automatically assign geographical coordinates to videos. Obtained results show how our multimodal approach improves the geocoding results when compared to methods that rely on a single modality (either textual or visual descriptors). We also show that the proposed multimodal approach yields comparable results to the best submissions to the Placing Task in 2012 using no extra information besides the available development/training data. Another contribution of this work is related to the proposal of a new effectiveness evaluation measure. The proposed measure is based on distance scores that summarize how effective a designed/tested approach is, considering its overall result for a test dataset. © 2013 Springer Science+Business Media New York.

Enhancing Activity Recognition by Fusing Inertial and Biometric Information

Activity recognition is an active research field nowadays, as it enables the development of highly adaptive applications, e.g. in the field of personal health. In this paper, a light high-level fusion algorithm to detect the activity that an individual is performing is presented. The algorithm relies on data gathered from accelerometers placed on different parts of the body, and on biometric sensors. Inertial sensors allow detecting activity by analyzing signal features such as amplitude or peaks. In addition, there is a relationship between the activity intensity and biometric response, which can be considered together with acceleration data to improve the accuracy of activity detection. The proposed algorithm is designed to work with minimum computational cost, being ready to run in a mobile device as part of a context-aware application. In order to enable different user scenarios, the algorithm offers best-effort activity estimation: its quality of estimation depends on the position and number of the available inertial sensors, and also on the presence of biometric information.

Brain ictal state characterisation through multimodal information integration

This Thesis addresses the problem of automated false-positive free detection of epileptic events by the fusion of information extracted from simultaneously recorded electro-encephalographic (EEG) and the electrocardiographic (ECG) time-series. The approach relies on a biomedical case for the coupling of the Brain and Heart systems through the central autonomic network during temporal lobe epileptic events: neurovegetative manifestations associated with temporal lobe epileptic events consist of alterations to the cardiac rhythm. From a neurophysiological perspective, epileptic episodes are characterised by a loss of complexity of the state of the brain. The description of arrhythmias, from a probabilistic perspective, observed during temporal lobe epileptic events and the description of the complexity of the state of the brain, from an information theory perspective, are integrated in a fusion-of-information framework towards temporal lobe epileptic seizure detection. The main contributions of the Thesis include the introduction of a biomedical case for the coupling of the Brain and Heart systems during temporal lobe epileptic seizures, partially reported in the clinical literature; the investigation of measures for the characterisation of ictal events from the EEG time series towards their integration in a fusion-of-knowledge framework; the probabilistic description of arrhythmias observed during temporal lobe epileptic events towards their integration in a fusion-of-knowledge framework; and the investigation of the different levels of the fusion-of-information architecture at which to perform the combination of information extracted from the EEG and ECG time-series. The performance of the method designed in the Thesis for the false-positive free automated detection of epileptic events achieved a false-positives rate of zero on the dataset of long-term recordings used in the Thesis.

Multimodal Mri-based Study In Patients With Spg4 Mutations.

Mutations in the SPG4 gene (SPG4-HSP) are the most frequent cause of hereditary spastic paraplegia, but the extent of the neurodegeneration related to the disease is not yet known. Therefore, our objective is to identify regions of the central nervous system damaged in patients with SPG4-HSP using a multi-modal neuroimaging approach. In addition, we aimed to identify possible clinical correlates of such damage. Eleven patients (mean age 46.0 ± 15.0 years, 8 men) with molecular confirmation of hereditary spastic paraplegia, and 23 matched healthy controls (mean age 51.4 ± 14.1years, 17 men) underwent MRI scans in a 3T scanner. We used 3D T1 images to perform volumetric measurements of the brain and spinal cord. We then performed tract-based spatial statistics and tractography analyses of diffusion tensor images to assess microstructural integrity of white matter tracts. Disease severity was quantified with the Spastic Paraplegia Rating Scale. Correlations were then carried out between MRI metrics and clinical data. Volumetric analyses did not identify macroscopic abnormalities in the brain of hereditary spastic paraplegia patients. In contrast, we found extensive fractional anisotropy reduction in the corticospinal tracts, cingulate gyri and splenium of the corpus callosum. Spinal cord morphometry identified atrophy without flattening in the group of patients with hereditary spastic paraplegia. Fractional anisotropy of the corpus callosum and pyramidal tracts did correlate with disease severity. Hereditary spastic paraplegia is characterized by relative sparing of the cortical mantle and remarkable damage to the distal portions of the corticospinal tracts, extending into the spinal cord.

Developmental Potential of Bovine Hand-Made Clone Embryos Reconstructed by Aggregation or Fusion with Distinct Cytoplasmic Volumes

Animal cloning has been associated with developmental abnormalities, with the level of heteroplasmy caused by the procedure being one of its potential limiting factors. The aim of this study was to determine the effect of the fusion of hemicytoplasts or aggregation of hemiembryos, varying the final cytoplasmic volume, on development and cell density of embryos produced by hand-made cloning (HMC), parthenogenesis or by in vitro fertilization (IVF). One or two enucleated hemicytoplasts were paired and fused with one skin somatic cell. Activated clone and zona-free parthenote embryos and hemiembryos were in vitro cultured in the well-of-the-well (WOW) system, being allocated to one of six experimental groups, on a per WOW basis: single clone or parthenote hemiembryos (1 x 50%); aggregation of two (2 x 50%), three (3 x 50%), or four (4 x 50%) clone or parthenote hemiembryos; single clone or parthenote embryos (1 x 100%); or aggregation of two clone or parthenote embryos (2 x 100%). Control zona-intact parthenote or IVF embryos were in vitro cultured in four-well dishes. Results indicated that the increase in the number of aggregated structures within each WOW was followed by a linear increase in cleavage, blastocyst rate, and cell density. The increase in cytoplasmic volume, either by fusion or by aggregation, had a positive effect on embryo development, supporting the establishment of pregnancies and the birth of a viable clone calf after transfer to recipients. However, embryo aggregation did not improve development on a hemicytoplast basis, except for the aggregation of two clone embryos.

Near-Barrier Fusion of the (8)B+(58)Ni Proton-Halo System

Fusion cross sections were measured for the exotic proton-halo nucleus (8)B incident on a (58)Ni target at several energies near the Coulomb barrier. This is the first experiment to report on the fusion of a protonhalo nucleus. The resulting excitation function shows a striking enhancement with respect to expectations for normal projectiles. Evidence is presented that the sum of the fusion and breakup yields saturates the total reaction cross section.