Multi-modal abnormality detection in video with unknown data segmentation

This paper examines a new problem in large scale stream data: abnormality detection which is localized to a data segmentation process. Unlike traditional abnormality detection methods which typically build one unified model across data stream, we propose that building multiple detection models focused on different coherent sections of the video stream would result in better detection performance. One key challenge is to segment the data into coherent sections as the number of segments is not known in advance and can vary greatly across cameras; and a principled way approach is required. To this end, we first employ the recently proposed infinite HMM and collapsed Gibbs inference to automatically infer data segmentation followed by constructing abnormality detection models which are localized to each segmentation. We demonstrate the superior performance of the proposed framework in a real-world surveillance camera data over 14 days.

Quantifying heteroskedasticity via binary decomposition

This paper presents a quantifying measure for heteroskedasticity of a time series. In this research, heteroskedasticity levels are measured by decomposing the examined time series recursively into homoskedastic segments. Each segment of the examined time series is decomposed into smaller segments if it tests positively to heteroskedasticity tests. The final quantified value of the heteroskedasticity level is the number of homoskedastic segments. The proposed measure is robust and detects heteroskedasticity in small average variance datasets.

A novel bio-kinematic encoder for human exercise representation and decomposition - Part 1 : Indexing and modelling

Current bio-kinematic encoders use velocity, acceleration and angular information to encode human exercises. However, in exercise physiology there is a need to distinguish between the shape of the trajectory and its execution dynamics. In this paper we propose such a two-component model and explore how best to compute these components of an action. In particular, we show how a new spatial indexing scheme, derived directly from the underlying differential geometry of curves, provides robust estimates of the shape and dynamics compared to standard temporal indexing schemes.

A novel bio-kinematic encoder for human exercise representation and decomposition - Part 2 : Robustness and Optimisation

Bio-kinematic characterisations of human exercises constitute dealing with parameters such as velocity, acceleration, joint angles, etc. A majority of these are measured directly from various sensors ranging from RGB cameras to inertial sensors. However, due to certain limitations associated with these sensors, such as inherent noise, filters are required to be implemented to subjugate the effect from the noise. When the two-component (trajectory shape and dynamics) bio-kinematic encoding model is being established to represent an exercise, reducing the effect from noise embedded in raw data will be important since the underlying model can be quite sensitive to noise. In this paper, we examine and compare some commonly used filters, namely least-square Gaussian filter, Savitzky-Golay filter and optimal Kalman filter, with four groups of real data collected from Microsoft Kinectc , and assert that Savitzky- Golay filter is the best one when establishing an underlying model for human exercise representation.

A system decomposition approach to the design of functional observers

This paper reports a system decomposition that allows the construction of a minimum-order functional observer using a state observer design approach. The system decomposition translates the functional observer design problem to that of a state observer for a smaller decomposed subsystem. Functional observability indices are introduced, and a closed-form expression for the minimum order required for a functional observer is derived in terms of those functional observability indices. © 2014 Taylor & Francis.

A spatial decomposition approach for investigating house price convergences

 Convergence of house prices indicates how prices are reaching an aggregate equilibrium in a long-run perspective. Identifying the convergence is important for cross-region housing development and investment. Few studies have identified house price convergences at different levels, with spatial effects on house prices predominantly ignored. The research presented here developed a spatial panel regression approach to investigate the convergences of house prices in Australian capital cities. Three hypotheses were tested to identify the level of house price convergence. The results demonstrate that a steady state in a system of regional house prices and spatial effects contribute to the convergence continuing.

Mixing vector construction for single channel semi-blind source separation using empirical mode decomposition

The Empirical Mode Decomposition (EMD) method is a commonly used method for solving the problem of single channel blind source separation (SCBSS) in signal processing. However, the mixing vector of SCBSS, which is the base of the EMD method, has not yet been effectively constructed. The mixing vector reflects the weights of original signal sources that form the single channel blind signal source. In this paper, we propose a novel method to construct a mixing vector for a single channel blind signal source to approximate the actual mixing vector in terms of keeping the same ratios between signal weights. The constructed mixing vector can be used to improve signal separations. Our method incorporates the adaptive filter, least square method, EMD method and signal source samples to construct the mixing vector. Experimental tests using audio signal evaluations were conducted and the results indicated that our method can improve the similar values of sources energy ratio from 0.2644 to 0.8366. This kind of recognition is very important in weak signal detection.

Effects of mucor mucedo on corncob decomposition in pyr-contaminated soil remediation

The effect of polycyclic aromatic hydrocarbon and highly effective degradation fungi Mucor mucedo (MU) was studied on corncob decomposition in Pyr-contaminated soil for 120 days to identify the impact of a degradable immobilized carrier on the remediation of soil contaminated by persistent organic pollutants. Results showed that the corncob was mainly composed of hemicelluloses, cellulose, and water dissolved (WD) material, which accounted for 85 percent of its total weight. MU addition significantly affected corncob decomposition. Thus, humic acid production and WD and benzene-ethanol dissolved material degradation increased. The peaking of the WD content was delayed for 30 days or more. The extractable pyrene content positively correlated with the WD content in the corncob during the decomposition. These results theoretically support a refined remediation principle of immobilized microorganisms.

Decomposition properties of PVA/graphene composites during melting-crystallization

Abstract The thermal decomposition of PVA and PVA composites during the melting-crystallization process is still unclear due to indistinct changes in chemical compositions. Using graphene as a model, the decomposition properties of PVA and PVA-graphene composites were systematically analyzed under multiple melting-crystallization cycles. And a series of isothermal decomposition experiments around the melting-crystallization temperature were carried out to simulate the corresponding decomposition kinetics. Based on multiple cycle melting-crystallization, the weight loss of PVA and PVA/graphene composites was successfully quantified. Further morphology investigation and chemical structure analysis indicated that the decomposition was non-uniformly distributed, rendering the possibility of crystallization for PVA and PVA/graphene composites after multiple heating-cooling cycles. In addition, isothermal decomposition analysis based on reduced time plot approach and model-free iso-conversional method indicated that Avrami-Eroffev model could well match the decomposition process of the neat PVA and PG-0.3 composite, while the Avrami-Eroffev and first order models could precisely forecast the decomposition of PG-0.9 composite. Both analyses during multiple cycle melting-crystallization and isothermal decomposition demonstrated that graphene served as decomposition accelerator in the whole thermal decomposition process, and particularly the decomposition of neat PVA and PVA/graphene composites was highly related to the band area ratios of C-H and O-H vibrations in Fourier transform infrared (FTIR) spectrum.