Detecting rare events using Kullback-Leibler divergence

One main challenge in developing a system for visual surveillance event detection is the annotation of target events in the training data. By making use of the assumption that events with security interest are often rare compared to regular behaviours, this paper presents a novel approach by using Kullback-Leibler (KL) divergence for rare event detection in a weakly supervised learning setting, where only clip-level annotation is available. It will be shown that this approach outperforms state-of-the-art methods on a popular real-world dataset, while preserving real time performance.

Detecting rare events using Kullback-Leibler divergence: A weakly supervised approach

Video surveillance infrastructure has been widely installed in public places for security purposes. However, live video feeds are typically monitored by human staff, making the detection of important events as they occur difficult. As such, an expert system that can automatically detect events of interest in surveillance footage is highly desirable. Although a number of approaches have been proposed, they have significant limitations: supervised approaches, which can detect a specific event, ideally require a large number of samples with the event spatially and temporally localised; while unsupervised approaches, which do not require this demanding annotation, can only detect whether an event is abnormal and not specific event types. To overcome these problems, we formulate a weakly-supervised approach using Kullback-Leibler (KL) divergence to detect rare events. The proposed approach leverages the sparse nature of the target events to its advantage, and we show that this data imbalance guarantees the existence of a decision boundary to separate samples that contain the target event from those that do not. This trait, combined with the coarse annotation used by weakly supervised learning (that only indicates approximately when an event occurs), greatly reduces the annotation burden while retaining the ability to detect specific events. Furthermore, the proposed classifier requires only a decision threshold, simplifying its use compared to other weakly supervised approaches. We show that the proposed approach outperforms state-of-the-art methods on a popular real-world traffic surveillance dataset, while preserving real time performance.

Automatic domain ontology extraction for context-sensitive opinion mining

Automated analysis of the sentiments presented in online consumer feedbacks can facilitate both organizations’ business strategy development and individual consumers’ comparison shopping. Nevertheless, existing opinion mining methods either adopt a context-free sentiment classification approach or rely on a large number of manually annotated training examples to perform context sensitive sentiment classification. Guided by the design science research methodology, we illustrate the design, development, and evaluation of a novel fuzzy domain ontology based contextsensitive opinion mining system. Our novel ontology extraction mechanism underpinned by a variant of Kullback-Leibler divergence can automatically acquire contextual sentiment knowledge across various product domains to improve the sentiment analysis processes. Evaluated based on a benchmark dataset and real consumer reviews collected from Amazon.com, our system shows remarkable performance improvement over the context-free baseline.

Learning domain-specific sentiment lexicons for predicting product sales

Generic sentiment lexicons have been widely used for sentiment analysis these days. However, manually constructing sentiment lexicons is very time-consuming and it may not be feasible for certain application domains where annotation expertise is not available. One contribution of this paper is the development of a statistical learning based computational method for the automatic construction of domain-specific sentiment lexicons to enhance cross-domain sentiment analysis. Our initial experiments show that the proposed methodology can automatically generate domain-specific sentiment lexicons which contribute to improve the effectiveness of opinion retrieval at the document level. Another contribution of our work is that we show the feasibility of applying the sentiment metric derived based on the automatically constructed sentiment lexicons to predict product sales of certain product categories. Our research contributes to the development of more effective sentiment analysis system to extract business intelligence from numerous opinionated expressions posted to the Web

Information-theoretic analysis of brain white matter fiber orientation distribution functions

We propose a new information-theoretic metric, the symmetric Kullback-Leibler divergence (sKL-divergence), to measure the difference between two water diffusivity profiles in high angular resolution diffusion imaging (HARDI). Water diffusivity profiles are modeled as probability density functions on the unit sphere, and the sKL-divergence is computed from a spherical harmonic series, which greatly reduces computational complexity. Adjustment of the orientation of diffusivity functions is essential when the image is being warped, so we propose a fast algorithm to determine the principal direction of diffusivity functions using principal component analysis (PCA). We compare sKL-divergence with other inner-product based cost functions using synthetic samples and real HARDI data, and show that the sKL-divergence is highly sensitive in detecting small differences between two diffusivity profiles and therefore shows promise for applications in the nonlinear registration and multisubject statistical analysis of HARDI data.

Fluid registration of diffusion tensor images using information theory

We apply an information-theoretic cost metric, the symmetrized Kullback-Leibler (sKL) divergence, or $J$-divergence, to fluid registration of diffusion tensor images. The difference between diffusion tensors is quantified based on the sKL-divergence of their associated probability density functions (PDFs). Three-dimensional DTI data from 34 subjects were fluidly registered to an optimized target image. To allow large image deformations but preserve image topology, we regularized the flow with a large-deformation diffeomorphic mapping based on the kinematics of a Navier-Stokes fluid. A driving force was developed to minimize the $J$-divergence between the deforming source and target diffusion functions, while reorienting the flowing tensors to preserve fiber topography. In initial experiments, we showed that the sKL-divergence based on full diffusion PDFs is adaptable to higher-order diffusion models, such as high angular resolution diffusion imaging (HARDI). The sKL-divergence was sensitive to subtle differences between two diffusivity profiles, showing promise for nonlinear registration applications and multisubject statistical analysis of HARDI data.

How does angular resolution affect diffusion imaging measures?

A key question in diffusion imaging is how many diffusion-weighted images suffice to provide adequate signal-to-noise ratio (SNR) for studies of fiber integrity. Motion, physiological effects, and scan duration all affect the achievable SNR in real brain images, making theoretical studies and simulations only partially useful. We therefore scanned 50 healthy adults with 105-gradient high-angular resolution diffusion imaging (HARDI) at 4T. From gradient image subsets of varying size (6 ≤ N ≤ 94) that optimized a spherical angular distribution energy, we created SNR plots (versus gradient numbers) for seven common diffusion anisotropy indices: fractional and relative anisotropy (FA, RA), mean diffusivity (MD), volume ratio (VR), geodesic anisotropy (GA), its hyperbolic tangent (tGA), and generalized fractional anisotropy (GFA). SNR, defined in a region of interest in the corpus callosum, was near-maximal with 58, 66, and 62 gradients for MD, FA, and RA, respectively, and with about 55 gradients for GA and tGA. For VR and GFA, SNR increased rapidly with more gradients. SNR was optimized when the ratio of diffusion-sensitized to non-sensitized images was 9.13 for GA and tGA, 10.57 for FA, 9.17 for RA, and 26 for MD and VR. In orientation density functions modeling the HARDI signal as a continuous mixture of tensors, the diffusion profile reconstruction accuracy rose rapidly with additional gradients. These plots may help in making trade-off decisions when designing diffusion imaging protocols.

Analyzing multi-fiber reconstruction in high angular resolution diffusion imaging using the tensor distribution function

High-angular resolution diffusion imaging (HARDI) can reconstruct fiber pathways in the brain with extraordinary detail, identifying anatomical features and connections not seen with conventional MRI. HARDI overcomes several limitations of standard diffusion tensor imaging, which fails to model diffusion correctly in regions where fibers cross or mix. As HARDI can accurately resolve sharp signal peaks in angular space where fibers cross, we studied how many gradients are required in practice to compute accurate orientation density functions, to better understand the tradeoff between longer scanning times and more angular precision. We computed orientation density functions analytically from tensor distribution functions (TDFs) which model the HARDI signal at each point as a unit-mass probability density on the 6D manifold of symmetric positive definite tensors. In simulated two-fiber systems with varying Rician noise, we assessed how many diffusionsensitized gradients were sufficient to (1) accurately resolve the diffusion profile, and (2) measure the exponential isotropy (EI), a TDF-derived measure of fiber integrity that exploits the full multidirectional HARDI signal. At lower SNR, the reconstruction accuracy, measured using the Kullback-Leibler divergence, rapidly increased with additional gradients, and EI estimation accuracy plateaued at around 70 gradients.

Modelling Cued-Target Recall using Quantum Inspired Models of Target Activation

This article presents and evaluates Quantum Inspired models of Target Activation using Cued-Target Recall Memory Modelling over multiple sources of Free Association data. Two components were evaluated: Whether Quantum Inspired models of Target Activation would provide a better framework than their classical psychological counterparts and how robust these models are across the different sources of Free Association data. In previous work, a formal model of cued-target recall did not exist and as such Target Activation was unable to be assessed directly. Further to that, the data source used was suspected of suffering from temporal and geographical bias. As a consequence, Target Activation was measured against cued-target recall data as an approximation of performance. Since then, a formal model of cued-target recall (PIER3) has been developed [10] with alternative sources of data also becoming available. This allowed us to directly model target activation in cued-target recall with human cued-target recall pairs and use multiply sources of Free Association Data. Featural Characteristics known to be important to Target Activation were measured for each of the data sources to identify any major differences that may explain variations in performance for each of the models. Each of the activation models were used in the PIER3 memory model for each of the data sources and was benchmarked against cued-target recall pairs provided by the University of South Florida (USF). Two methods where used to evaluate performance. The first involved measuring the divergence between the sets of results using the Kullback Leibler (KL) divergence with the second utilizing a previous statistical analysis of the errors [9]. Of the three sources of data, two were sourced from human subjects being the USF Free Association Norms and the University of Leuven (UL) Free Association Networks. The third was sourced from a new method put forward by Galea and Bruza, 2015 in which pseudo Free Association Networks (Corpus Based Association Networks - CANs) are built using co-occurrence statistics on large text corpus. It was found that the Quantum Inspired Models of Target Activation not only outperformed the classical psychological model but was more robust across a variety of data sources.

Modeling strategies in longitudinal data analysis: Covariate, variance function and correlation structure selection

A modeling paradigm is proposed for covariate, variance and working correlation structure selection for longitudinal data analysis. Appropriate selection of covariates is pertinent to correct variance modeling and selecting the appropriate covariates and variance function is vital to correlation structure selection. This leads to a stepwise model selection procedure that deploys a combination of different model selection criteria. Although these criteria find a common theoretical root based on approximating the Kullback-Leibler distance, they are designed to address different aspects of model selection and have different merits and limitations. For example, the extended quasi-likelihood information criterion (EQIC) with a covariance penalty performs well for covariate selection even when the working variance function is misspecified, but EQIC contains little information on correlation structures. The proposed model selection strategies are outlined and a Monte Carlo assessment of their finite sample properties is reported. Two longitudinal studies are used for illustration.