Robust speaker verification via fusion of speech and lip modalities

This paper investigates the use of lip information, in conjunction with speech information, for robust speaker verification in the presence of background noise. It has been previously shown in our own work, and in the work of others, that features extracted from a speaker's moving lips hold speaker dependencies which are complementary with speech features. We demonstrate that the fusion of lip and speech information allows for a highly robust speaker verification system which outperforms the performance of either sub-system. We present a new technique for determining the weighting to be applied to each modality so as to optimize the performance of the fused system. Given a correct weighting, lip information is shown to be highly effective for reducing the false acceptance and false rejection error rates in the presence of background noise

The use of speech and lip modalities for robust speaker verification under adverse conditions

Investigates the use of lip information, in conjunction with speech information, for robust speaker verification in the presence of background noise. We have previously shown (Int. Conf. on Acoustics, Speech and Signal Proc., vol. 6, pp. 3693-3696, May 1998) that features extracted from a speaker's moving lips hold speaker dependencies which are complementary with speech features. We demonstrate that the fusion of lip and speech information allows for a highly robust speaker verification system which outperforms either subsystem individually. We present a new technique for determining the weighting to be applied to each modality so as to optimize the performance of the fused system. Given a correct weighting, lip information is shown to be highly effective for reducing the false acceptance and false rejection error rates in the presence of background noise

Sequential decision fusion of multibiometrics applied to text-dependent speaker verification for controlled errors

Reliability of the performance of biometric identity verification systems remains a significant challenge. Individual biometric samples of the same person (identity class) are not identical at each presentation and performance degradation arises from intra-class variability and inter-class similarity. These limitations lead to false accepts and false rejects that are dependent. It is therefore difficult to reduce the rate of one type of error without increasing the other. The focus of this dissertation is to investigate a method based on classifier fusion techniques to better control the trade-off between the verification errors using text-dependent speaker verification as the test platform. A sequential classifier fusion architecture that integrates multi-instance and multisample fusion schemes is proposed. This fusion method enables a controlled trade-off between false alarms and false rejects. For statistically independent classifier decisions, analytical expressions for each type of verification error are derived using base classifier performances. As this assumption may not be always valid, these expressions are modified to incorporate the correlation between statistically dependent decisions from clients and impostors. The architecture is empirically evaluated by applying the proposed architecture for text dependent speaker verification using the Hidden Markov Model based digit dependent speaker models in each stage with multiple attempts for each digit utterance. The trade-off between the verification errors is controlled using the parameters, number of decision stages (instances) and the number of attempts at each decision stage (samples), fine-tuned on evaluation/tune set. The statistical validation of the derived expressions for error estimates is evaluated on test data. The performance of the sequential method is further demonstrated to depend on the order of the combination of digits (instances) and the nature of repetitive attempts (samples). The false rejection and false acceptance rates for proposed fusion are estimated using the base classifier performances, the variance in correlation between classifier decisions and the sequence of classifiers with favourable dependence selected using the 'Sequential Error Ratio' criteria. The error rates are better estimated by incorporating user-dependent (such as speaker-dependent thresholds and speaker-specific digit combinations) and class-dependent (such as clientimpostor dependent favourable combinations and class-error based threshold estimation) information. The proposed architecture is desirable in most of the speaker verification applications such as remote authentication, telephone and internet shopping applications. The tuning of parameters - the number of instances and samples - serve both the security and user convenience requirements of speaker-specific verification. The architecture investigated here is applicable to verification using other biometric modalities such as handwriting, fingerprints and key strokes.

Architecture research and hardware implementation on simplified neural computing system for face identification

This paper describes a special-purpose neural computing system for face identification. The system architecture and hardware implementation are introduced in detail. An algorithm based on biomimetic pattern recognition has been embedded. For the total 1200 tests for face identification, the false rejection rate is 3.7% and the false acceptance rate is 0.7%.

Targeted Impersonation As A Tool For The Detection Of Biometric System Vulnerabilities

This paper argues that biometric verification evaluations can obscure vulnerabilities that increase the chances that an attacker could be falsely accepted. This can occur because existing evaluations implicitly assume that an imposter claiming a false identity would claim a random identity rather than consciously selecting a target to impersonate. This paper shows how an attacker can select a target with a similar biometric signature in order to increase their chances of false acceptance. It demonstrates this effect using a publicly available iris recognition algorithm. The evaluation shows that the system can be vulnerable to attackers targeting subjects who are enrolled with a smaller section of iris due to occlusion. The evaluation shows how the traditional DET curve analysis conceals this vulnerability. As a result, traditional analysis underestimates the importance of an existing score normalisation method for addressing occlusion. The paper concludes by evaluating how the targeted false acceptance rate increases with the number of available targets. Consistent with a previous investigation of targeted face verification performance, the experiment shows that the false acceptance rate can be modelled using the traditional FAR measure with an additional term that is proportional to the logarithm of the number of available targets.

TARGETED BIOMETRIC IMPERSONATION

When applying biometric algorithms to forensic verification, false acceptance and false rejection can mean a failure to identify a criminal, or worse, lead to the prosecution of individuals for crimes they did not commit. It is therefore critical that biometric evaluations be performed as accurately as possible to determine their legitimacy as a forensic tool. This paper argues that, for forensic verification scenarios, traditional performance measures are insufficiently accurate. This inaccuracy occurs because existing verification evaluations implicitly assume that an imposter claiming a false identity would claim a random identity rather than consciously selecting a target to impersonate. In addition to describing this new vulnerability, the paper describes a novel Targeted.. FAR metric that combines the traditional False Acceptance Rate (FAR) measure with a term that indicates how performance degrades with the number of potential targets. The paper includes an evaluation of the effects of targeted impersonation on an existing academic face verification system. This evaluation reveals that even with a relatively small number of targets false acceptance rates can increase significantly, making the analysed biometric systems unreliable.

Investigation into DCT Feature Selection for Visual Lip-Based Biometric Authentication

This paper investigated using lip movements as a behavioural biometric for person authentication. The system was trained, evaluated and tested using the XM2VTS dataset, following the Lausanne Protocol configuration II. Features were selected from the DCT coefficients of the greyscale lip image. This paper investigated the number of DCT coefficients selected, the selection process, and static and dynamic feature combinations. Using a Gaussian Mixture Model - Universal Background Model framework an Equal Error Rate of 2.20% was achieved during evaluation and on an unseen test set a False Acceptance Rate of 1.7% and False Rejection Rate of 3.0% was achieved. This compares favourably with face authentication results on the same dataset whilst not being susceptible to spoofing attacks.

Towards hand biometrics in mobile devices

The increasing demand of security oriented to mobile applications has raised the attention to biometrics, as a proper and suitable solution for providing secure environment to mobile devices. With this aim, this document presents a biometric system based on hand geometry oriented to mobile devices, involving a high degree of freedom in terms of illumination, hand rotation and distance to camera. The user takes a picture of their own hand in the free space, without requiring any flat surface to locate the hand, and without removals of rings, bracelets or watches. The proposed biometric system relies on an accurate segmentation procedure, able to isolate hands from any background; a feature extraction, invariant to orientation, illumination, distance to camera and background; and a user classification, based on k-Nearest Neighbor approach, able to provide an accurate results on individual identification. The proposed method has been evaluated with two own databases collected with a HTC mobile. First database contains 120 individuals, with 20 acquisitions of both hands. Second database is a synthetic database, containing 408000 images of hand samples in different backgrounds: tiles, grass, water, sand, soil and the like. The system is able to identify individuals properly with False Reject Rate of 5.78% and False Acceptance Rate of 0.089%, using 60 features (15 features per finger)

A high efficient biometrics approach for unconstrained iris segmentation and recognition

This dissertation develops an innovative approach towards less-constrained iris biometrics. Two major contributions are made in this research endeavor: (1) Designed an award-winning segmentation algorithm in the less-constrained environment where image acquisition is made of subjects on the move and taken under visible lighting conditions, and (2) Developed a pioneering iris biometrics method coupling segmentation and recognition of the iris based on video of moving persons under different acquisitions scenarios. The first part of the dissertation introduces a robust and fast segmentation approach using still images contained in the UBIRIS (version 2) noisy iris database. The results show accuracy estimated at 98% when using 500 randomly selected images from the UBIRIS.v2 partial database, and estimated at 97% in a Noisy Iris Challenge Evaluation (NICE.I) in an international competition that involved 97 participants worldwide involving 35 countries, ranking this research group in sixth position. This accuracy is achieved with a processing speed nearing real time. The second part of this dissertation presents an innovative segmentation and recognition approach using video-based iris images. Following the segmentation stage which delineates the iris region through a novel segmentation strategy, some pioneering experiments on the recognition stage of the less-constrained video iris biometrics have been accomplished. In the video-based and less-constrained iris recognition, the test or subject iris videos/images and the enrolled iris images are acquired with different acquisition systems. In the matching step, the verification/identification result was accomplished by comparing the similarity distance of encoded signature from test images with each of the signature dataset from the enrolled iris images. With the improvements gained, the results proved to be highly accurate under the unconstrained environment which is more challenging. This has led to a false acceptance rate (FAR) of 0% and a false rejection rate (FRR) of 17.64% for 85 tested users with 305 test images from the video, which shows great promise and high practical implications for iris biometrics research and system design.

Do multiple true-false items beat the commonly used one-best-answers questions regarding to the Ottawa Criteria for Good Assessment? Results of a literature review

Background: Multiple True-False-Items (MTF-Items) might offer some advantages compared to one-best-answer-questions (TypeA) as they allow more than one correct answer and may better represent clinical decisions. However, in medical education assessment MTF-Items are seldom used. Summary of Work: With this literature review existing findings on MTF-items and on TypeA were compared along the Ottawa Criteria for Good Assessment, i.e. (1) reproducibility, (2) feasibility, (3) validity, (4) acceptance, (5) educational effect, (6) catalytic effects, and (7) equivalence. We conducted a literature research on ERIC and Google Scholar including papers from the years 1935 to 2014. We used the search terms “multiple true-false”, “true-false”, “true/false”, and “Kprim” combined with “exam”, “test”, and “assessment”. Summary of Results: We included 29 out of 33 studies. Four of them were carried out in the medical field Compared to TypeA, MTF-Items are associated with (1) higher reproducibility (2) lower feasibility (3) similar validity (4) higher acceptance (5) higher educational effect (6) no studies on catalytic effects or (7) equivalence. Discussion and Conclusions: While studies show overall good characteristics of MTF items according to the Ottawa criteria, this type of question seems to be rather seldom used. One reason might be the reported lower feasibility. Overall the literature base is still weak. Furthermore, only 14 % of literature is from the medical domain. Further studies to better understand the characteristics of MTF-Items in the medical domain are warranted. Take-home messages: Overall the literature base is weak and therefore further studies are needed. Existing studies show that: MTF-Items show higher reliability, acceptance and educational effect; MTF-Items are more difficult to produce

Psychological and Social Perspectives on Malingering and Symptom Exaggeration: Conceptualization and Interventions Using Acceptance and Commitment Therapy

Malingering and the production of false symptoms seen in such disorders as Factitious Disorder are an ongoing mystery to medical and mental health professionals. Historically, these presentations have been difficult to identify and treat. As might be expected, individuals with such symptomology rarely agree to participate in research, possibly because of a reluctance to admit to the feigning or exaggerating behaviors and a fear of reprisals. Many different etiologies have been proposed, including the assumption of roles in order to manage impressions, taking control of symptoms in order to gain attention or other rewards or avoid aversive events, and even the production of symptoms that is largely out of awareness such as is seen in conversion or somatoform presentations. By examining historical and present-day beliefs about etiology and treatment interventions, professionals can explore what new types of effective treatment might look like. The behaviorist philosophy that underlies Acceptance and Commitment Therapy proposes a perspective emphasizing effective working in context. This philosophy also suggests individuals sometimes engage in behavior in order to escape from or avoid aversive experiences. Utilizing case examples and fresh behavioral perspectives provides insight and ideas for conceptualization of these behaviors of interest. Using the above conceptualizations, an ACT based treatment of those who produce false symptoms is introduced.

Rape myth in true and false rape allegations

Police records of 38 rape allegations, evenly split into maintained-as-true and withdrawn-as-false categories were compared with 19 generated-false statements from recruited participants. The Illinois Rape Myth Acceptance Scale (IRMAS) was used to assess the attitudes of the participants and a content analysis derived from IRMAS was used to compare the three categories of allegation. Rape myths were present in all three allegation types. The two categories of false allegation both contained more rape myths than the true allegations but no differences were found between the generated and withdrawn false allegations. High scorers in IRMAS also produced more violent false accounts. In addition to these findings, this study provides support for the further examination of rape myths in both false and true statements and use of generated allegations as proxies for real false statements.

The effect of young adult literature on adolescents' rape myth acceptance

This quasi-experimental study (N = 139) measured the effect of a reader response based instructional unit of the novel Speak on adolescents' rape myth acceptance. Participants were eighth grade language arts students at a Title I middle school in a major metropolitan school district. Seven classes were randomly assigned to treatment ( n = 4) or control (n = 3) condition. Two teachers participated in the study and both taught both treatment and control classes. The study lasted a period of five weeks. Participants were pretested using the Rape Myth Acceptance Scale (Burt, 1980) and a researcher created scale, the Adolescent Date Rape Scale (ADRMS). Analysis of pretests showed the ADRMS to be a reliable and valid measure of rape myth acceptance in adolescents. Factor analysis revealed it to have two major components: "She Wanted It" and "She Lied." Pretests supported previous studies which found girls to have significantly lower initial levels of rape myth acceptance than boys (p < .001). A 2 (group) x 2 (instructor) x 2 (sex) ANCOVA using ADRMS pretest as a covariate and ADRMS posttest as a dependent variable found that treatment was effective in reducing rape myth acceptance (p < .001, η2 = .15). Boys with high rape myth acceptance as demonstrated by pretest scores of 1 standard deviation above the mean on ADRMS did not have a backlash to treatment. Extended analysis revealed that participants had significantly lower scores posttest on Factor 1, "She Wanted It" (p < .001, η2 = .27), while scores on Factor 2, "She Lied" were not significantly lower (p = .07). This may be because the content of the novel primarily deals with issues questioning whether the main characters assault was a rape rather than a false accusation. Attrition rates were low (N = 15) and attrition analysis showed that drop outs did not significantly alter the treatment or control groups. Implications for reader response instruction of young adult literature, for research on rape myth acceptance in secondary schools, and for statistical analysis of effect size using pretests as filters are discussed.