Classifier selection using sequential error ratio criterion for multi-instance and multi-sample fusion

Classifier selection is a problem encountered by multi-biometric systems that aim to improve performance through fusion of decisions. A particular decision fusion architecture that combines multiple instances (n classifiers) and multiple samples (m attempts at each classifier) has been proposed in previous work to achieve controlled trade-off between false alarms and false rejects. Although analysis on text-dependent speaker verification has demonstrated better performance for fusion of decisions with favourable dependence compared to statistically independent decisions, the performance is not always optimal. Given a pool of instances, best performance with this architecture is obtained for certain combination of instances. Heuristic rules and diversity measures have been commonly used for classifier selection but it is shown that optimal performance is achieved for the `best combination performance' rule. As the search complexity for this rule increases exponentially with the addition of classifiers, a measure - the sequential error ratio (SER) - is proposed in this work that is specifically adapted to the characteristics of sequential fusion architecture. The proposed measure can be used to select a classifier that is most likely to produce a correct decision at each stage. Error rates for fusion of text-dependent HMM based speaker models using SER are compared with other classifier selection methodologies. SER is shown to achieve near optimal performance for sequential fusion of multiple instances with or without the use of multiple samples. The methodology applies to multiple speech utterances for telephone or internet based access control and to other systems such as multiple finger print and multiple handwriting sample based identity verification systems.

Multi-level information fusion for environment aware robotic navigation

This thesis develops the hardware and software framework for an integrated navigation system. Dynamic data fusion algorithms are used to develop a system with a high level of resistance to the typical problems that affect standard navigation systems.

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.

Brain-inspired sensor fusion for navigating robots

Current state of the art robot mapping and navigation systems produce impressive performance under a narrow range of robot platform, sensor and environmental conditions, in contrast to animals such as rats that produce “good enough” maps that enable them to function under an incredible range of situations. In this paper we present a rat-inspired featureless sensor-fusion system that assesses the usefulness of multiple sensor modalities based on their utility and coherence for place recognition, without knowledge as to the type of sensor. We demonstrate the system on a Pioneer robot in indoor and outdoor environments with abrupt lighting changes. Through dynamic weighting of the sensors, the system is able to perform correct place recognition and mapping where the static sensor weighting approach fails.

Decision fusion from parts and samples for robust iris recognition

Fusion techniques can be used in biometrics to achieve higher accuracy. When biometric systems are in operation and the threat level changes, controlling the trade-off between detection error rates can reduce the impact of an attack. In a fused system, varying a single threshold does not allow this to be achieved, but systematic adjustment of a set of parameters does. In this paper, fused decisions from a multi-part, multi-sample sequential architecture are investigated for that purpose in an iris recognition system. A specific implementation of the multi-part architecture is proposed and the effect of the number of parts and samples in the resultant detection error rate is analysed. The effectiveness of the proposed architecture is then evaluated under two specific cases of obfuscation attack: miosis and mydriasis. Results show that robustness to such obfuscation attacks is achieved, since lower error rates than in the case of the non-fused base system are obtained.

Laser-radar data fusion with Gaussian Process Implicit Surfaces

This work considers the problem of building high-fidelity 3D representations of the environment from sensor data acquired by mobile robots. Multi-sensor data fusion allows for more complete and accurate representations, and for more reliable perception, especially when different sensing modalities are used. In this paper, we propose a thorough experimental analysis of the performance of 3D surface reconstruction from laser and mm-wave radar data using Gaussian Process Implicit Surfaces (GPIS), in a realistic field robotics scenario. We first analyse the performance of GPIS using raw laser data alone and raw radar data alone, respectively, with different choices of covariance matrices and different resolutions of the input data. We then evaluate and compare the performance of two different GPIS fusion approaches. The first, state-of-the-art approach directly fuses raw data from laser and radar. The alternative approach proposed in this paper first computes an initial estimate of the surface from each single source of data, and then fuses these two estimates. We show that this method outperforms the state of the art, especially in situations where the sensors react differently to the targets they perceive.

Book review: ‘Anterior Eye Disease and Therapeutics A-Z' [2nd ed.] Edition, by A. Bruce and M. Loughnan. Churchill Livingstone (Elsevier), Edinburgh 2011

With the ever-increasing emphasis on ocular disease recognition in the practice of optometry and especially anterior eye disease management and therapeutics, any book addressing such issues is bound to have a captive audience. This second edition of Anterior Eye Disease and Therapeutics A–Z provides a succinct yet comprehensive coverage of this topic.

Autonomous multisensor calibration and closed-loop fusion for SLAM

The vast majority of current robot mapping and navigation systems require specific well-characterized sensors that may require human-supervised calibration and are applicable only in one type of environment. Furthermore, if a sensor degrades in performance, either through damage to itself or changes in environmental conditions, the effect on the mapping system is usually catastrophic. In contrast, the natural world presents robust, reasonably well-characterized solutions to these problems. Using simple movement behaviors and neural learning mechanisms, rats calibrate their sensors for mapping and navigation in an incredibly diverse range of environments and then go on to adapt to sensor damage and changes in the environment over the course of their lifetimes. In this paper, we introduce similar movement-based autonomous calibration techniques that calibrate place recognition and self-motion processes as well as methods for online multisensor weighting and fusion. We present calibration and mapping results from multiple robot platforms and multisensory configurations in an office building, university campus, and forest. With moderate assumptions and almost no prior knowledge of the robot, sensor suite, or environment, the methods enable the bio-inspired RatSLAM system to generate topologically correct maps in the majority of experiments.

An acute growth factor treatment that preserves function after spinal cord contusion injury

Inflammation of the spinal cord after traumatic spinal cord injury leads to destruction of healthy tissue. This “secondary degeneration” is more damaging than the initial physical damage and is the major contributor to permanent loss of functions. In our previous study we showed that combined delivery of two growth factors, vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF), significantly reduced secondary degeneration after hemi-section injury of the spinal cord in the rat. Growth factor treatment reduced the size of the lesion cavity at 30d compared to control animals and further reduced the cavity at 90d in treated animals while in control animals the lesion cavity continued to increase in size. Growth factor treatment also reduced astrogliosis and reduced macroglia/macrophage activation around the injury site. Treatment with individual growth factors alone had similar effects to control treatments. The present study investigated whether growth factor treatment would improve locomotor behaviour after spinal contusion injury, a more relevant preclinical model of spinal cord injury. The growth factors were delivered for the first 7d to the injury site via osmotic minipump. Locomotor behaviour was monitored at 1-28d after injury using the BBB score and at 30d using automated gait analysis. Treated animals had BBB scores of 18; Control animals scored 10. Treated animals had significantly reduced lesion cavities and reduced macroglia/macrophage activation around the injury site. We conclude that growth factor treatment preserved spinal cord tissues after contusion injury, thereby allowing functional recovery. This treatment has the potential to significantly reduce the severity of human spinal cord injuries.

Thoracoscopic scoliosis surgery – a cheaper option? A cost analysis of 189 consecutive cases

INTRODUCTION: Increasing health care costs, limited resources and increased demand makes cost effective and cost-efficient delivery of Adolescent Idiopathic Scoliosis (AIS) management paramount. Rising implant costs in deformity correction surgery have prompted analysis of whether high implant densities are justified. The objective of this study was to analyse the costs of thoracoscopic scoliosis surgery, comparing initial learning curve costs with those of the established technique and to the costs involved in posterior instrumented fusion from the literature. METHODS: 189 consecutive cases from April 2000 to July 2011 were assessed with a minimum of 2 years follow-up. Information was gathered from a prospective database covering perioperative factors, clinical and radiological outcomes, complications and patient reported outcomes. The patients were divided into three groups to allow comparison; 1. A learning curve cohort, 2. An intermediate cohort and 3. A third cohort of patients, using our established technique. Hospital finance records and implant manufacturer figures were corrected to 2013 costs. A literature review of AIS management costs and implant density in similar curve types was performed. RESULTS: The mean pre-op Cobb angle was 53°(95%CI 0.4) and was corrected postop to mean 22.9°(CI 0.4). The overall complication rate was 20.6%, primarily in the first cohort, with a rate of 5.6% in the third cohort. The average total costs were $46,732, operating room costs of $10,301 (22.0%) and ICU costs of $4620 (9.8%). The mean number of screws placed was 7.1 (CI 0.04) with a single rod used for each case giving average implant costs of $14,004 (29.9%). Comparison of the three groups revealed higher implant costs as the technique evolved to that in use today, from $13,049 in Group 1 to $14577 in Group 3 (P<0.001). Conversely operating room costs reduced from $10,621 in Group 1 to $7573 (P<0.001) in Group 3. ICU stay was reduced from an average of 1.2 to 0 days. In-patient stay was significantly (P=0.006) lower in Groups 2 and 3 (5.4 days) than Group 1 (5.9 days) (i.e. a reduction in cost of approximately $6,140). CONCLUSIONS: The evolution of our thoracoscopic anterior scoliosis correction has resulted in an increase in the number of levels fused and reduction in complication rate. Implant costs have risen as a result, however, there has been a concurrent decrease in those costs generated by operating room use, ICU and in-patient stay with increasing experience. Literature review of equivalent curve types treated posteriorly shows similar perioperative factors but higher implant density, 69-83% compared to the 50% in this study. Thoracoscopic Scoliosis surgery presents a low density, reliable, efficient and effective option for selected curves. A cost analysis of Thoracoscopic Scoliosis Surgery using financial records and a prospectively collected database of all patients since 2000, demonstrating a clear cost advantage compared to equivalent posterior instrumentation and fusion.

Anterior temporal cortex and semantic memory : reconciling findings from neuropsychology and functional imaging

Studies of semantic impairment arising from brain disease suggest that the anterior temporal lobes are critical for semantic abilities in humans; yet activation of these regions is rarely reported in functional imaging studies of healthy controls performing semantic tasks. Here, we combined neuropsychological and PET functional imaging data to show that when healthy subjects identify concepts at a specific level, the regions activated correspond to the site of maximal atrophy in patients with relatively pure semantic impairment. The stimuli were color photographs of common animals or vehicles, and the task was category verification at specific (e.g., robin), intermediate (e.g., bird), or general (e.g., animal) levels. Specific, relative to general, categorization activated the antero-lateral temporal cortices bilaterally, despite matching of these experimental conditions for difficulty. Critically, in patients with atrophy in precisely these areas, the most pronounced deficit was in the retrieval of specific semantic information.

Using patient specific finite element models to understand the biomechanics of paediatric spinal deformity surgery

Adolescent idiopathic scoliosis (AIS) is a spinal deformity, which may require surgical correction by attaching rods to the patient’s spine using screws inserted into the vertebrae. Complication rates for deformity correction surgery are unacceptably high. Determining an achievable correction without overloading the adjacent spinal tissues or implants requires an understanding of the mechanical interaction between these components. We have developed novel patient specific modelling software to create individualized finite element models (FEM) representing the thoracolumbar spine and ribcage of scoliosis patients. We are using these models to better understand the biomechanics of spinal deformity correction.

Robust multiple-sensing-modality data fusion using Gaussian Process Implicit Surfaces

The ability to build high-fidelity 3D representations of the environment from sensor data is critical for autonomous robots. Multi-sensor data fusion allows for more complete and accurate representations. Furthermore, using distinct sensing modalities (i.e. sensors using a different physical process and/or operating at different electromagnetic frequencies) usually leads to more reliable perception, especially in challenging environments, as modalities may complement each other. However, they may react differently to certain materials or environmental conditions, leading to catastrophic fusion. In this paper, we propose a new method to reliably fuse data from multiple sensing modalities, including in situations where they detect different targets. We first compute distinct continuous surface representations for each sensing modality, with uncertainty, using Gaussian Process Implicit Surfaces (GPIS). Second, we perform a local consistency test between these representations, to separate consistent data (i.e. data corresponding to the detection of the same target by the sensors) from inconsistent data. The consistent data can then be fused together, using another GPIS process, and the rest of the data can be combined as appropriate. The approach is first validated using synthetic data. We then demonstrate its benefit using a mobile robot, equipped with a laser scanner and a radar, which operates in an outdoor environment in the presence of large clouds of airborne dust and smoke.