SAIVT-ADMRG @ MediaEval 2014 social event detection

This paper outlines the approach taken by the Speech, Audio, Image and Video Technologies laboratory, and the Applied Data Mining Research Group (SAIVT-ADMRG) in the 2014 MediaEval Social Event Detection (SED) task. We participated in the event based clustering subtask (subtask 1), and focused on investigating the incorporation of image features as another source of data to aid clustering. In particular, we developed a descriptor based around the use of super-pixel segmentation, that allows a low dimensional feature that incorporates both colour and texture information to be extracted and used within the popular bag-of-visual-words (BoVW) approach.

Representing team behaviours from noisy data using player role

Due to their unobtrusive nature, vision-based approaches to tracking sports players have been preferred over wearable sensors as they do not require the players to be instrumented for each match. Unfortunately however, due to the heavy occlusion between players, variation in resolution and pose, in addition to fluctuating illumination conditions, tracking players continuously is still an unsolved vision problem. For tasks like clustering and retrieval, having noisy data (i.e. missing and false player detections) is problematic as it generates discontinuities in the input data stream. One method of circumventing this issue is to use an occupancy map, where the field is discretised into a series of zones and a count of player detections in each zone is obtained. A series of frames can then be concatenated to represent a set-play or example of team behaviour. A problem with this approach though is that the compressibility is low (i.e. the variability in the feature space is incredibly high). In this paper, we propose the use of a bilinear spatiotemporal basis model using a role representation to clean-up the noisy detections which operates in a low-dimensional space. To evaluate our approach, we used a fully instrumented field-hockey pitch with 8 fixed high-definition (HD) cameras and evaluated our approach on approximately 200,000 frames of data from a state-of-the-art real-time player detector and compare it to manually labeled data.

Multi-action recognition via stochastic modelling of optical flow and gradients

In this paper we propose a novel approach to multi-action recognition that performs joint segmentation and classification. This approach models each action using a Gaussian mixture using robust low-dimensional action features. Segmentation is achieved by performing classification on overlapping temporal windows, which are then merged to produce the final result. This approach is considerably less complicated than previous methods which use dynamic programming or computationally expensive hidden Markov models (HMMs). Initial experiments on a stitched version of the KTH dataset show that the proposed approach achieves an accuracy of 78.3%, outperforming a recent HMM-based approach which obtained 71.2%.

Linear programming for large-scale Markov decision problems

We consider the problem of controlling a Markov decision process (MDP) with a large state space, so as to minimize average cost. Since it is intractable to compete with the optimal policy for large scale problems, we pursue the more modest goal of competing with a low-dimensional family of policies. We use the dual linear programming formulation of the MDP average cost problem, in which the variable is a stationary distribution over state-action pairs, and we consider a neighborhood of a low-dimensional subset of the set of stationary distributions (defined in terms of state-action features) as the comparison class. We propose a technique based on stochastic convex optimization and give bounds that show that the performance of our algorithm approaches the best achievable by any policy in the comparison class. Most importantly, this result depends on the size of the comparison class, but not on the size of the state space. Preliminary experiments show the effectiveness of the proposed algorithm in a queuing application.

Fostering collaboration in the management of family finances

Managing finances is a practice carried out daily in homes across the world. Despite this, the practice is not yet a strong focus for HCI work in the home. This paper looks specifically at the ways that families organise and manage their finances. Based on a process of in-situ qualitative interviews, we find that money management is often a collaborative process in the family and occurs in a similar way to other household tasks; as part of existing divisions of domestic labour. Participating families reflected upon the creative and often low-tech systems they used to manage their finances and ideas they had for “ideal” technologies that would better help them organise their money. This paper presents ideas for fostering collaboration around family finances. Design propositions are made towards a Family Holiday Expense Tracker and family-situated bill payment reminder, both of which aim to encourage whole-of-family collaboration around money management and better including children in what is, at-present, an activity they are largely excluded from.

Unsupervised Domain Adaptation by Domain Invariant Projection

Domain-invariant representations are key to addressing the domain shift problem where the training and test exam- ples follow different distributions. Existing techniques that have attempted to match the distributions of the source and target domains typically compare these distributions in the original feature space. This space, however, may not be di- rectly suitable for such a comparison, since some of the fea- tures may have been distorted by the domain shift, or may be domain specific. In this paper, we introduce a Domain Invariant Projection approach: An unsupervised domain adaptation method that overcomes this issue by extracting the information that is invariant across the source and tar- get domains. More specifically, we learn a projection of the data to a low-dimensional latent space where the distance between the empirical distributions of the source and target examples is minimized. We demonstrate the effectiveness of our approach on the task of visual object recognition and show that it outperforms state-of-the-art methods on a stan- dard domain adaptation benchmark dataset

Low-complexity estimation of CFO and frequency independent I/Q mismatch for OFDM systems

CFO and I/Q mismatch could cause significant performance degradation to OFDM systems. Their estimation and compensation are generally difficult as they are entangled in the received signal. In this paper, we propose some low-complexity estimation and compensation schemes in the receiver, which are robust to various CFO and I/Q mismatch values although the performance is slightly degraded for very small CFO. These schemes consist of three steps: forming a cosine estimator free of I/Q mismatch interference, estimating I/Q mismatch using the estimated cosine value, and forming a sine estimator using samples after I/Q mismatch compensation. These estimators are based on the perception that an estimate of cosine serves much better as the basis for I/Q mismatch estimation than the estimate of CFO derived from the cosine function. Simulation results show that the proposed schemes can improve system performance significantly, and they are robust to CFO and I/Q mismatch.

Three-dimensional printing of hierarchical and tough mesoporous bioactive glass scaffolds with a controllable pore architecture, excellent mechanical strength and mineralization ability

New-generation biomaterials for bone regenerations should be highly bioactive, resorbable and mechanically strong. Mesoporous bioactive glass (MBG), as a novel bioactive material, has been used for the study of bone regeneration due to its excellent bioactivity, degradation and drug-delivery ability; however, how to construct a 3D MBG scaffold (including other bioactive inorganic scaffolds) for bone regeneration still maintains a significant challenge due to its/their inherit brittleness and low strength. In this brief communication, we reported a new facile method to prepare hierarchical and multifunctional MBG scaffolds with controllable pore architecture, excellent mechanical strength and mineralization ability for bone regeneration application by a modified 3D-printing technique using polyvinylalcohol (PVA), as a binder. The method provides a new way to solve the commonly existing issues for inorganic scaffold materials, for example, uncontrollable pore architecture, low strength, high brittleness and the requirement for the second sintering at high temperature. The obtained 3D-printing MBG scaffolds possess a high mechanical strength which is about 200 times for that of traditional polyurethane foam template-resulted MBG scaffolds. They have highly controllable pore architecture, excellent apatite-mineralization ability and sustained drug-delivery property. Our study indicates that the 3D-printed MBG scaffolds may be an excellent candidate for bone regeneration.

Pattern recognition using invariants defined from higher order spectra - one-dimensional inputs

A new approach to pattern recognition using invariant parameters based on higher order spectra is presented. In particular, invariant parameters derived from the bispectrum are used to classify one-dimensional shapes. The bispectrum, which is translation invariant, is integrated along straight lines passing through the origin in bifrequency space. The phase of the integrated bispectrum is shown to be scale and amplification invariant, as well. A minimal set of these invariants is selected as the feature vector for pattern classification, and a minimum distance classifier using a statistical distance measure is used to classify test patterns. The classification technique is shown to distinguish two similar, but different bolts given their one-dimensional profiles. Pattern recognition using higher order spectral invariants is fast, suited for parallel implementation, and has high immunity to additive Gaussian noise. Simulation results show very high classification accuracy, even for low signal-to-noise ratios.

A comparison of two commercially available motion capture systems for gait analysis : high end vs low-cost

In 1999 Richards compared the accuracy of commercially available motion capture systems commonly used in biomechanics. Richards identified that in static tests the optical motion capture systems generally produced RMS errors of less than 1.0 mm. During dynamic tests, the RMS error increased to up to 4.2 mm in some systems. In the last 12 years motion capture systems have continued to evolve and now include high-resolution CCD or CMOS image sensors, wireless communication, and high full frame sampling frequencies. In addition to hardware advances, there have also been a number of advances in software, which includes improved calibration and tracking algorithms, real time data streaming, and the introduction of the c3d standard. These advances have allowed the system manufactures to maintain a high retail price in the name of advancement. In areas such as gait analysis and ergonomics many of the advanced features such as high resolution image sensors and high sampling frequencies are not required due to the nature of the task often investigated. Recently Natural Point introduced low cost cameras, which on face value appear to be suitable as at very least a high quality teaching tool in biomechanics and possibly even a research tool when coupled with the correct calibration and tracking software. The aim of the study was therefore to compare both the linear accuracy and quality of angular kinematics from a typical high end motion capture system and a low cost system during a simple task.

Specification & design of safety technologies for complex socio-technical systems : low-cost level crossings case study

The introduction of safety technologies into complex socio-technical systems requires an integrated and holistic approach to HF and engineering, considering the effects of failures not only within system boundaries, but also at the interfaces with other systems and humans. Level crossing warning devices are examples of such systems where technically safe states within the system boundary can influence road user performance, giving rise to other hazards that degrade safety of the system. Chris will discuss the challenges that have been encountered to date in developing a safety argument in support of low-cost level crossing warning devices. The design and failure modes of level crossing warning devices are known to have a significant influence on road user performance; however, quantifying this effect is one of the ongoing challenges in determining appropriate reliability and availability targets for low-cost level crossing warning devices.

A low-complexity flight controller for Unmanned Aircraft Systems with constrained control allocation

In this paper, we propose a framework for joint allocation and constrained control design of flight controllers for Unmanned Aircraft Systems (UAS). The actuator configuration is used to map actuator constraint set into the space of the aircraft generalised forces. By constraining the demanded generalised forces, we ensure that the allocation problem is always feasible; and therefore, it can be solved without constraints. This leads to an allocation problem that does not require on-line numerical optimisation. Furthermore, since the controller handles the constraints, and there is no need to implement heuristics to inform the controller about actuator saturation. The latter is fundamental for avoiding Pilot Induced Oscillations (PIO) in remotely operated UAS due to the rate limit on the aircraft control surfaces.

Diagnostics and two-dimensional simulation of low-frequency inductively coupled plasmas with neutral gas heating and electron heat fluxes

This article presents the results on the diagnostics and numerical modeling of low-frequency (∼460 KHz) inductively coupled plasmas generated in a cylindrical metal chamber by an external flat spiral coil. Experimental data on the electron number densities and temperatures, electron energy distribution functions, and optical emission intensities of the abundant plasma species in low/intermediate pressure argon discharges are included. The spatial profiles of the plasma density, electron temperature, and excited argon species are computed, for different rf powers and working gas pressures, using the two-dimensional fluid approach. The model allows one to achieve a reasonable agreement between the computed and experimental data. The effect of the neutral gas temperature on the plasma parameters is also investigated. It is shown that neutral gas heating (at rf powers≥0.55kW) is one of the key factors that control the electron number density and temperature. The dependence of the average rf power loss, per electron-ion pair created, on the working gas pressure shows that the electron heat flux to the walls appears to be a critical factor in the total power loss in the discharge.

A dynamic navigation model for unmanned aircraft systems and an application to autonomous front-on environmental sensing and photography using low-cost sensor systems

This paper presents an unmanned aircraft system (UAS) that uses a probabilistic model for autonomous front-on environmental sensing or photography of a target. The system is based on low-cost and readily-available sensor systems in dynamic environments and with the general intent of improving the capabilities of dynamic waypoint-based navigation systems for a low-cost UAS. The behavioural dynamics of target movement for the design of a Kalman filter and Markov model-based prediction algorithm are included. Geometrical concepts and the Haversine formula are applied to the maximum likelihood case in order to make a prediction regarding a future state of a target, thus delivering a new waypoint for autonomous navigation. The results of the application to aerial filming with low-cost UAS are presented, achieving the desired goal of maintained front-on perspective without significant constraint to the route or pace of target movement.

Impacts of wind power integration on small signal stability and low frequency oscillation characteristics of interconnected power systems

为研究风电并网对互联系统低频振荡的影响,基于完整的双馈风电机组模型,定性分析了两区域互联系统在风电机组并网前后阻尼特性的变化情况.从双馈风电机组并网输送距离、并网容量、互联系统联络线传送功率、是否加装电力系统稳定器等多个方面,多角度分析了风电场并网对互联系统小干扰稳定及低频振荡特性的影响.之后,以两个包括两个区域的电力系统为例,进行了系统的计算分析和比较.结果表明,有双馈风电机组接入的互联电力系统,在不同运行模式下,双馈风电机组的并网输送距离、出力水平、联络线传送功率对低频振荡模式的影响在趋势和程度上均有显著差异,这样在对风电场进行入网规划、设计和运行时就需要综合考虑这些因素的影响.