Conrad Sanderson

Majid Namazi, M. A. Hakim Newton, Conrad Sanderson et al.

A travelling thief problem (TTP) is a proxy to real-life problems such as postal collection. TTP comprises an entanglement of a travelling salesman problem (TSP) and a knapsack problem (KP) since items of KP are scattered over cities of TSP, and a thief has to visit cities to collect items. In TTP, city selection and item selection decisions need close coordination since the thief's travelling speed depends on the knapsack's weight and the order of visiting cities affects the order of item collection. Existing TTP solvers deal with city selection and item selection separately, keeping decisions for one type unchanged while dealing with the other type. This separation essentially means very poor coordination between two types of decision. In this paper, we first show that a simple local search based coordination approach does not work in TTP. Then, to address the aforementioned problems, we propose a human designed coordination heuristic that makes changes to collection plans during exploration of cyclic tours. We further propose another human designed coordination heuristic that explicitly exploits the cyclic tours in item selections during collection plan exploration. Lastly, we propose a machine learning based coordination heuristic that captures characteristics of the two human designed coordination heuristics. Our proposed coordination based approaches help our TTP solver significantly outperform existing state-of-the-art TTP solvers on a set of benchmark problems. Our solver is named Cooperation Coordination (CoCo) and its source code is available from https://github.com/majid75/CoCo

Joel Janek Dabrowski, Daniel Edward Pagendam, James Hilton et al.

We apply the Physics Informed Neural Network (PINN) to the problem of wildfire fire-front modelling. We use the PINN to solve the level-set equation, which is a partial differential equation that models a fire-front through the zero-level-set of a level-set function. The result is a PINN that simulates a fire-front as it propagates through the spatio-temporal domain. We show that popular optimisation cost functions used in the literature can result in PINNs that fail to maintain temporal continuity in modelled fire-fronts when there are extreme changes in exogenous forcing variables such as wind direction. We thus propose novel additions to the optimisation cost function that improves temporal continuity under these extreme changes. Furthermore, we develop an approach to perform data assimilation within the PINN such that the PINN predictions are drawn towards observations of the fire-front. Finally, we incorporate our novel approaches into a Bayesian PINN (B-PINN) to provide uncertainty quantification in the fire-front predictions. This is significant as the standard solver, the level-set method, does not naturally offer the capability for data assimilation and uncertainty quantification. Our results show that, with our novel approaches, the B-PINN can produce accurate predictions with high quality uncertainty quantification on real-world data.

Conrad Sanderson, David Douglas, Qinghua Lu

Many sets of ethics principles for responsible AI have been proposed to allay concerns about misuse and abuse of AI/ML systems. The underlying aspects of such sets of principles include privacy, accuracy, fairness, robustness, explainability, and transparency. However, there are potential tensions between these aspects that pose difficulties for AI/ML developers seeking to follow these principles. For example, increasing the accuracy of an AI/ML system may reduce its explainability. As part of the ongoing effort to operationalise the principles into practice, in this work we compile and discuss a catalogue of 10 notable tensions, trade-offs and other interactions between the underlying aspects. We primarily focus on two-sided interactions, drawing on support spread across a diverse literature. This catalogue can be helpful in raising awareness of the possible interactions between aspects of ethics principles, as well as facilitating well-supported judgements by the designers and developers of AI/ML systems.

Andrew Bolt, Carolyn Huston, Petra Kuhnert et al.

Computational simulations of wildfire spread typically employ empirical rate-of-spread calculations under various conditions (such as terrain, fuel type, weather). Small perturbations in conditions can often lead to significant changes in fire spread (such as speed and direction), necessitating a computationally expensive large set of simulations to quantify uncertainty. Model emulation seeks alternative representations of physical models using machine learning, aiming to provide more efficient and/or simplified surrogate models. We propose a dedicated spatio-temporal neural network based framework for model emulation, able to capture the complex behaviour of fire spread models. The proposed approach can approximate forecasts at fine spatial and temporal resolutions that are often challenging for neural network based approaches. Furthermore, the proposed approach is robust even with small training sets, due to novel data augmentation methods. Empirical experiments show good agreement between simulated and emulated firefronts, with an average Jaccard score of 0.76.

Ryan R. Curtin, Marcus Edel, Conrad Sanderson

We describe the Bandicoot GPU linear algebra toolkit, a C++ based library that prioritises ease of use without compromising efficiency. Bandicoot's API is compatible with the popular Armadillo CPU linear algebra library, enabling easy transition for existing CPU-based codebases. Unlike other GPU-focused toolkits, Bandicoot uses template metaprogramming to generate fused GPU kernels directly at compile time, yielding efficient kernels that are often able to saturate memory bandwidth. This removes the need for runtime overhead or JIT infrastructure. Empirical results show that Bandicoot outperforms (sometimes by considerable margins) commonly-used linear algebra toolkits including PyTorch, TensorFlow, and JAX.

Sevvandi Kandanaarachchi, Mahdi Abolghasemi, Hideya Ochiai et al.

Continuously evolving cyber-attacks against industrial networks reduce the effectiveness of signature-based detection methods. Once malware has infiltrated a network (for example, entering via an unsecured device), it can infect further network nodes and carry out malicious activity. Infected nodes can exhibit unusual behaviour in their use of Address Resolution Protocol (ARP) calls within the network. In order to detect such anomalous nodes, we propose a two-stage method: (i) modelling of ARP call behaviour via hierarchical time series prediction methods, and (ii) exploiting Extreme Value Theory (EVT) to robustly detect whether deviations from expected behaviour are anomalous. EVT is able to handle heavy-tailed distributions which are exhibited by internet traffic. Empirical evaluations on a real-life dataset containing over 10M ARP calls from 362 nodes show that the proposed method results in considerably reduced number of false positives, addressing the problem of alert fatigue commonly reported by security professionals.

Conrad Sanderson, Ryan Curtin

Modelling of multivariate densities is a core component in many signal processing, pattern recognition and machine learning applications. The modelling is often done via Gaussian mixture models (GMMs), which use computationally expensive and potentially unstable training algorithms. We provide an overview of a fast and robust implementation of GMMs in the C++ language, employing multi-threaded versions of the Expectation Maximisation (EM) and k-means training algorithms. Multi-threading is achieved through reformulation of the EM and k-means algorithms into a MapReduce-like framework. Furthermore, the implementation uses several techniques to improve numerical stability and modelling accuracy. We demonstrate that the multi-threaded implementation achieves a speedup of an order of magnitude on a recent 16 core machine, and that it can achieve higher modelling accuracy than a previously well-established publically accessible implementation. The multi-threaded implementation is included as a user-friendly class in recent releases of the open source Armadillo C++ linear algebra library. The library is provided under the permissive Apache~2.0 license, allowing unencumbered use in commercial products.

Sevvandi Kandanaarachchi, Ziqi Xu, Stefan Westerlund et al.

Many dynamic processes such as telecommunication and transport networks can be described through discrete time series of graphs. Modelling the dynamics of such time series enables prediction of graph structure at future time steps, which can be used in applications such as detection of anomalies. Existing approaches for graph prediction have limitations such as assuming that the vertices do not to change between consecutive graphs. To address this, we propose to exploit time series prediction methods in combination with an adapted form of flux balance analysis (FBA), a linear programming method originating from biochemistry. FBA is adapted to incorporate various constraints applicable to the scenario of growing graphs. Empirical evaluations on synthetic datasets (constructed via Preferential Attachment model) and real datasets (UCI Message, HePH, Facebook, Bitcoin) demonstrate the efficacy of the proposed approach.

Ben Harwood, Amir Dezfouli, Iadine Chades et al.

Approximate k-Nearest Neighbour (ANN) methods are often used for mining information and aiding machine learning on large scale high-dimensional datasets. ANN methods typically differ in the index structure used for accelerating searches, resulting in various recall/runtime trade-off points. For applications with static datasets, runtime constraints and dataset properties can be used to empirically select an ANN method with suitable operating characteristics. However, for applications with dynamic datasets, which are subject to frequent online changes (like addition of new samples), there is currently no consensus as to which ANN methods are most suitable. Traditional evaluation approaches do not consider the computational costs of updating the index structure, as well as the rate and size of index updates. To address this, we empirically evaluate 5 popular ANN methods on two main applications (online data collection and online feature learning) while taking into account these considerations. Two dynamic datasets are used, derived from the SIFT1M dataset with 1 million samples and the DEEP1B dataset with 1 billion samples. The results indicate that the often used k-d trees method is not suitable on dynamic datasets as it is slower than a straightforward baseline exhaustive search method. For online data collection, the Hierarchical Navigable Small World Graphs method achieves a consistent speedup over baseline across a wide range of recall rates. For online feature learning, the Scalable Nearest Neighbours method is faster than baseline for recall rates below 75%.

Conrad Sanderson, Emma Schleiger, David Douglas et al.

While the operationalisation of high-level AI ethics principles into practical AI/ML systems has made progress, there is still a theory-practice gap in managing tensions between the underlying AI ethics aspects. We cover five approaches for addressing the tensions via trade-offs, ranging from rudimentary to complex. The approaches differ in the types of considered context, scope, methods for measuring contexts, and degree of justification. None of the approaches is likely to be appropriate for all organisations, systems, or applications. To address this, we propose a framework which consists of: (i) proactive identification of tensions, (ii) prioritisation and weighting of ethics aspects, (iii) justification and documentation of trade-off decisions. The proposed framework aims to facilitate the implementation of well-rounded AI/ML systems that are appropriate for potential regulatory requirements.

Andrew Bolt, Conrad Sanderson, Joel Janek Dabrowski et al.

Wildfire propagation is a highly stochastic process where small changes in environmental conditions (such as wind speed and direction) can lead to large changes in observed behaviour. A traditional approach to quantify uncertainty in fire-front progression is to generate probability maps via ensembles of simulations. However, use of ensembles is typically computationally expensive, which can limit the scope of uncertainty analysis. To address this, we explore the use of a spatio-temporal neural-based modelling approach to directly estimate the likelihood of fire propagation given uncertainty in input parameters. The uncertainty is represented by deliberately perturbing the input weather forecast during model training. The computational load is concentrated in the model training process, which allows larger probability spaces to be explored during deployment. Empirical evaluations indicate that the proposed model achieves comparable fire boundaries to those produced by the traditional SPARK simulation platform, with an overall Jaccard index (similarity score) of 67.4% on a set of 35 simulated fires. When compared to a related neural model (emulator) which was employed to generate probability maps via ensembles of emulated fires, the proposed approach produces competitive Jaccard similarity scores while being approximately an order of magnitude faster.

Conrad Sanderson, David Douglas, Qinghua Lu et al.

As consensus across the various published AI ethics principles is approached, a gap remains between high-level principles and practical techniques that can be readily adopted to design and develop responsible AI systems. We examine the practices and experiences of researchers and engineers from Australia's national scientific research agency (CSIRO), who are involved in designing and developing AI systems for many application areas. Semi-structured interviews were used to examine how the practices of the participants relate to and align with a set of high-level AI ethics principles proposed by the Australian Government. The principles comprise: (1) privacy protection and security, (2) reliability and safety, (3) transparency and explainability, (4) fairness, (5) contestability, (6) accountability, (7) human-centred values, (8) human, social and environmental wellbeing. Discussions on the gained insights from the interviews include various tensions and trade-offs between the principles, and provide suggestions for implementing each high-level principle. We also present suggestions aiming to enhance associated support mechanisms.

Qinghua Lu, Liming Zhu, Xiwei Xu et al.

Although artificial intelligence (AI) is solving real-world challenges and transforming industries, there are serious concerns about its ability to behave and make decisions in a responsible way. Many AI ethics principles and guidelines for responsible AI have been recently issued by governments, organisations, and enterprises. However, these AI ethics principles and guidelines are typically high-level and do not provide concrete guidance on how to design and develop responsible AI systems. To address this shortcoming, we first present an empirical study where we interviewed 21 scientists and engineers to understand the practitioners' perceptions on AI ethics principles and their implementation. We then propose a template that enables AI ethics principles to be operationalised in the form of concrete patterns and suggest a list of patterns using the newly created template. These patterns provide concrete, operationalised guidance that facilitate the development of responsible AI systems.

Ryan R. Curtin, Marcus Edel, Rahul Ganesh Prabhu et al.

We overview the ensmallen numerical optimization library, which provides a flexible C++ framework for mathematical optimization of user-supplied objective functions. Many types of objective functions are supported, including general, differentiable, separable, constrained, and categorical. A diverse set of pre-built optimizers is provided, including Quasi-Newton optimizers and many variants of Stochastic Gradient Descent. The underlying framework facilitates the implementation of new optimizers. Optimization of an objective function typically requires supplying only one or two C++ functions. Custom behavior can be easily specified via callback functions. Empirical comparisons show that ensmallen outperforms other frameworks while providing more functionality. The library is available at https://ensmallen.org and is distributed under the permissive BSD license.

Conrad Sanderson, Dan Pagendam, Brendan Power et al.

With the underlying aim of increasing efficiency of computational modelling pertinent for managing & protecting the Great Barrier Reef, we perform a preliminary investigation on the use of deep neural networks for opportunistic model emulation of APSIM models by repurposing an existing large dataset containing outputs of APSIM model runs. The dataset has not been specifically tailored for the model emulation task. We employ two neural network architectures for the emulation task: densely connected feed-forward neural network (FFNN), and gated recurrent unit feeding into FFNN (GRU-FFNN), a type of a recurrent neural network. Various configurations of the architectures are trialled. A minimum correlation statistic is used to identify clusters of APSIM scenarios that can be aggregated to form training sets for model emulation. We focus on emulating 4 important outputs of the APSIM model: runoff, soil_loss, DINrunoff, Nleached. The GRU-FFNN architecture with three hidden layers and 128 units per layer provides good emulation of runoff and DINrunoff. However, soil_loss and Nleached were emulated relatively poorly under a wide range of the considered architectures; the emulators failed to capture variability at higher values of these two outputs. While the opportunistic data available from past modelling activities provides a large and useful dataset for exploring APSIM emulation, it may not be sufficiently rich enough for successful deep learning of more complex model dynamics. Design of Computer Experiments may be required to generate more informative data to emulate all output variables of interest. We also suggest the use of synthetic meteorology settings to allow the model to be fed a wide range of inputs. These need not all be representative of normal conditions, but can provide a denser, more informative dataset from which complex relationships between input and outputs can be learned.

Xiaolong Zhu, Fernando Vanegas, Felipe Gonzalez et al.

The use of multi-rotor Unmanned Aerial Vehicles (UAVs) for search and rescue as well as remote sensing is rapidly increasing. Multi-rotor UAVs, however, have limited endurance. The range of UAV applications can be widened if teams of multiple UAVs are used. We propose a framework for a team of UAVs to cooperatively explore and find a target in complex GPS-denied environments with obstacles. The team of UAVs autonomously navigates, explores, detects, and finds the target in a cluttered environment with a known map. Examples of such environments include indoor scenarios, urban or natural canyons, caves, and tunnels, where the GPS signal is limited or blocked. The framework is based on a probabilistic decentralised Partially Observable Markov Decision Process which accounts for the uncertainties in sensing and the environment. The team can cooperate efficiently, with each UAV sharing only limited processed observations and their locations during the mission. The system is simulated using the Robotic Operating System and Gazebo. Performance of the system with an increasing number of UAVs in several indoor scenarios with obstacles is tested. Results indicate that the proposed multi-UAV system has improvements in terms of time-cost, the proportion of search area surveyed, as well as successful rates for search and rescue missions.

Majid Namazi, Conrad Sanderson, M. A. Hakim Newton et al.

The travelling thief problem (TTP) is a multi-component optimisation problem involving two interdependent NP-hard components: the travelling salesman problem (TSP) and the knapsack problem (KP). Recent state-of-the-art TTP solvers modify the underlying TSP and KP solutions in an iterative and interleaved fashion. The TSP solution (cyclic tour) is typically changed in a deterministic way, while changes to the KP solution typically involve a random search, effectively resulting in a quasi-meandering exploration of the TTP solution space. Once a plateau is reached, the iterative search of the TTP solution space is restarted by using a new initial TSP tour. We propose to make the search more efficient through an adaptive surrogate model (based on a customised form of Support Vector Regression) that learns the characteristics of initial TSP tours that lead to good TTP solutions. The model is used to filter out non-promising initial TSP tours, in effect reducing the amount of time spent to find a good TTP solution. Experiments on a broad range of benchmark TTP instances indicate that the proposed approach filters out a considerable number of non-promising initial tours, at the cost of omitting only a small number of the best TTP solutions.

Ryan R. Curtin, Marcus Edel, Rahul Ganesh Prabhu et al.

This report provides an introduction to the ensmallen numerical optimization library, as well as a deep dive into the technical details of how it works. The library provides a fast and flexible C++ framework for mathematical optimization of arbitrary user-supplied functions. A large set of pre-built optimizers is provided, including many variants of Stochastic Gradient Descent and Quasi-Newton optimizers. Several types of objective functions are supported, including differentiable, separable, constrained, and categorical objective functions. Implementation of a new optimizer requires only one method, while a new objective function requires typically only one or two C++ methods. Through internal use of C++ template metaprogramming, ensmallen provides support for arbitrary user-supplied callbacks and automatic inference of unsupplied methods without any runtime overhead. Empirical comparisons show that ensmallen outperforms other optimization frameworks (such as Julia and SciPy), sometimes by large margins. The library is available at https://ensmallen.org and is distributed under the permissive BSD license.

Majid Namazi, Conrad Sanderson, M. A. Hakim Newton et al.

The travelling thief problem (TTP) is a representative of multi-component optimisation problems with interacting components. TTP combines the knapsack problem (KP) and the travelling salesman problem (TSP). A thief performs a cyclic tour through a set of cities, and pursuant to a collection plan, collects a subset of items into a rented knapsack with finite capacity. The aim is to maximise profit while minimising renting cost. Existing TTP solvers typically solve the KP and TSP components in an interleaved manner: the solution of one component is kept fixed while the solution of the other component is modified. This suggests low coordination between solving the two components, possibly leading to low quality TTP solutions. The 2-OPT heuristic is often used for solving the TSP component, which reverses a segment in the tour. Within TTP, 2-OPT does not take into account the collection plan, which can result in a lower objective value. This in turn can result in the tour modification to be rejected by a solver. We propose an expanded form of 2-OPT to change the collection plan in coordination with tour modification. Items regarded as less profitable and collected in cities located earlier in the reversed segment are substituted by items that tend to be more profitable and not collected in cities located later in the reversed segment. The collection plan is further changed through a modified form of the hill-climbing bit-flip search, where changes in the collection state are only permitted for boundary items, which are defined as lowest profitable collected items or highest profitable uncollected items. This restriction reduces the time spent on the KP component, allowing more tours to be evaluated by the TSP component within a time budget. The proposed approaches form the basis of a new cooperative coordination solver, which is shown to outperform several state-of-the-art TTP solvers.

Shikhar Bhardwaj, Ryan R. Curtin, Marcus Edel et al.

We present ensmallen, a fast and flexible C++ library for mathematical optimization of arbitrary user-supplied functions, which can be applied to many machine learning problems. Several types of optimizations are supported, including differentiable, separable, constrained, and categorical objective functions. The library provides many pre-built optimizers (including numerous variants of SGD and Quasi-Newton optimizers) as well as a flexible framework for implementing new optimizers and objective functions. Implementation of a new optimizer requires only one method and a new objective function requires typically one or two C++ functions. This can aid in the quick implementation and prototyping of new machine learning algorithms. Due to the use of C++ template metaprogramming, ensmallen is able to support compiler optimizations that provide fast runtimes. Empirical comparisons show that ensmallen is able to outperform other optimization frameworks (like Julia and SciPy), sometimes by large margins. The library is distributed under the BSD license and is ready for use in production environments.

Majid Namazi, Conrad Sanderson, M. A. Hakim Newton et al.

The well-known Late Acceptance Hill Climbing (LAHC) search aims to overcome the main downside of traditional Hill Climbing (HC) search, which is often quickly trapped in a local optimum due to strictly accepting only non-worsening moves within each iteration. In contrast, LAHC also accepts worsening moves, by keeping a circular array of fitness values of previously visited solutions and comparing the fitness values of candidate solutions against the least recent element in the array. While this straightforward strategy has proven effective, there are nevertheless situations where LAHC can unfortunately behave in a similar manner to HC. For example, when a new local optimum is found, often the same fitness value is stored many times in the array. To address this shortcoming, we propose new acceptance and replacement strategies to take into account worsening, improving, and sideways movement scenarios with the aim to improve the diversity of values in the array. Compared to LAHC, the proposed Diversified Late Acceptance Search approach is shown to lead to better quality solutions that are obtained with a lower number of iterations on benchmark Travelling Salesman Problems and Quadratic Assignment Problems.

ZongYuan Ge, Chris McCool, Conrad Sanderson et al.

Fine-grained classification is a relatively new field that has concentrated on using information from a single image, while ignoring the enormous potential of using video data to improve classification. In this work we present the novel task of video-based fine-grained object classification, propose a corresponding new video dataset, and perform a systematic study of several recent deep convolutional neural network (DCNN) based approaches, which we specifically adapt to the task. We evaluate three-dimensional DCNNs, two-stream DCNNs, and bilinear DCNNs. Two forms of the two-stream approach are used, where spatial and temporal data from two independent DCNNs are fused either via early fusion (combination of the fully-connected layers) and late fusion (concatenation of the softmax outputs of the DCNNs). For bilinear DCNNs, information from the convolutional layers of the spatial and temporal DCNNs is combined via local co-occurrences. We then fuse the bilinear DCNN and early fusion of the two-stream approach to combine the spatial and temporal information at the local and global level (Spatio-Temporal Co-occurrence). Using the new and challenging video dataset of birds, classification performance is improved from 23.1% (using single images) to 41.1% when using the Spatio-Temporal Co-occurrence system. Incorporating automatically detected bounding box location further improves the classification accuracy to 53.6%.

Johanna Carvajal, Arnold Wiliem, Conrad Sanderson et al.

Can we predict the winner of Miss Universe after watching how they stride down the catwalk during the evening gown competition? Fashion gurus say they can! In our work, we study this question from the perspective of computer vision. In particular, we want to understand whether existing computer vision approaches can be used to automatically extract the qualities exhibited by the Miss Universe winners during their catwalk. This study can pave the way towards new vision-based applications for the fashion industry. To this end, we propose a novel video dataset, called the Miss Universe dataset, comprising 10 years of the evening gown competition selected between 1996-2010. We further propose two ranking-related problems: (1) Miss Universe Listwise Ranking and (2) Miss Universe Pairwise Ranking. In addition, we also develop an approach that simultaneously addresses the two proposed problems. To describe the videos we employ the recently proposed Stacked Fisher Vectors in conjunction with robust local spatio-temporal features. From our evaluation we found that although the addressed problems are extremely challenging, the proposed system is able to rank the winner in the top 3 best predicted scores for 5 out of 10 Miss Universe competitions.

Johanna Carvajal, Chris McCool, Brian Lovell et al.

We propose a hierarchical approach to multi-action recognition that performs joint classification and segmentation. A given video (containing several consecutive actions) is processed via a sequence of overlapping temporal windows. Each frame in a temporal window is represented through selective low-level spatio-temporal features which efficiently capture relevant local dynamics. Features from each window are represented as a Fisher vector, which captures first and second order statistics. Instead of directly classifying each Fisher vector, it is converted into a vector of class probabilities. The final classification decision for each frame is then obtained by integrating the class probabilities at the frame level, which exploits the overlapping of the temporal windows. Experiments were performed on two datasets: s-KTH (a stitched version of the KTH dataset to simulate multi-actions), and the challenging CMU-MMAC dataset. On s-KTH, the proposed approach achieves an accuracy of 85.0%, significantly outperforming two recent approaches based on GMMs and HMMs which obtained 78.3% and 71.2%, respectively. On CMU-MMAC, the proposed approach achieves an accuracy of 40.9%, outperforming the GMM and HMM approaches which obtained 33.7% and 38.4%, respectively. Furthermore, the proposed system is on average 40 times faster than the GMM based approach.

Johanna Carvajal, Arnold Wiliem, Chris McCool et al.

We present a comparative evaluation of various techniques for action recognition while keeping as many variables as possible controlled. We employ two categories of Riemannian manifolds: symmetric positive definite matrices and linear subspaces. For both categories we use their corresponding nearest neighbour classifiers, kernels, and recent kernelised sparse representations. We compare against traditional action recognition techniques based on Gaussian mixture models and Fisher vectors (FVs). We evaluate these action recognition techniques under ideal conditions, as well as their sensitivity in more challenging conditions (variations in scale and translation). Despite recent advancements for handling manifolds, manifold based techniques obtain the lowest performance and their kernel representations are more unstable in the presence of challenging conditions. The FV approach obtains the highest accuracy under ideal conditions. Moreover, FV best deals with moderate scale and translation changes.

ZongYuan Ge, Alex Bewley, Christopher McCool et al.

We present a novel deep convolutional neural network (DCNN) system for fine-grained image classification, called a mixture of DCNNs (MixDCNN). The fine-grained image classification problem is characterised by large intra-class variations and small inter-class variations. To overcome these problems our proposed MixDCNN system partitions images into K subsets of similar images and learns an expert DCNN for each subset. The output from each of the K DCNNs is combined to form a single classification decision. In contrast to previous techniques, we provide a formulation to perform joint end-to-end training of the K DCNNs simultaneously. Extensive experiments, on three datasets using two network structures (AlexNet and GoogLeNet), show that the proposed MixDCNN system consistently outperforms other methods. It provides a relative improvement of 12.7% and achieves state-of-the-art results on two datasets.

Zongyuan Ge, Christopher Mccool, Conrad Sanderson et al.

Fine-grained categorisation has been a challenging problem due to small inter-class variation, large intra-class variation and low number of training images. We propose a learning system which first clusters visually similar classes and then learns deep convolutional neural network features specific to each subset. Experiments on the popular fine-grained Caltech-UCSD bird dataset show that the proposed method outperforms recent fine-grained categorisation methods under the most difficult setting: no bounding boxes are presented at test time. It achieves a mean accuracy of 77.5%, compared to the previous best performance of 73.2%. We also show that progressive transfer learning allows us to first learn domain-generic features (for bird classification) which can then be adapted to specific set of bird classes, yielding improvements in accuracy.

ZongYuan Ge, Chris McCool, Conrad Sanderson et al.

We propose a local modelling approach using deep convolutional neural networks (CNNs) for fine-grained image classification. Recently, deep CNNs trained from large datasets have considerably improved the performance of object recognition. However, to date there has been limited work using these deep CNNs as local feature extractors. This partly stems from CNNs having internal representations which are high dimensional, thereby making such representations difficult to model using stochastic models. To overcome this issue, we propose to reduce the dimensionality of one of the internal fully connected layers, in conjunction with layer-restricted retraining to avoid retraining the entire network. The distribution of low-dimensional features obtained from the modified layer is then modelled using a Gaussian mixture model. Comparative experiments show that considerable performance improvements can be achieved on the challenging Fish and UEC FOOD-100 datasets.

Johanna Carvajal, Conrad Sanderson, Chris McCool et al.

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%.

Mehrtash Harandi, Richard Hartley, Brian Lovell et al.

This paper introduces sparse coding and dictionary learning for Symmetric Positive Definite (SPD) matrices, which are often used in machine learning, computer vision and related areas. Unlike traditional sparse coding schemes that work in vector spaces, in this paper we discuss how SPD matrices can be described by sparse combination of dictionary atoms, where the atoms are also SPD matrices. We propose to seek sparse coding by embedding the space of SPD matrices into Hilbert spaces through two types of Bregman matrix divergences. This not only leads to an efficient way of performing sparse coding, but also an online and iterative scheme for dictionary learning. We apply the proposed methods to several computer vision tasks where images are represented by region covariance matrices. Our proposed algorithms outperform state-of-the-art methods on a wide range of classification tasks, including face recognition, action recognition, material classification and texture categorization.

Sareh Shirazi, Conrad Sanderson, Chris McCool et al.

We propose an adaptive tracking algorithm where the object is modelled as a continuously updated bag of affine subspaces, with each subspace constructed from the object's appearance over several consecutive frames. In contrast to linear subspaces, affine subspaces explicitly model the origin of subspaces. Furthermore, instead of using a brittle point-to-subspace distance during the search for the object in a new frame, we propose to use a subspace-to-subspace distance by representing candidate image areas also as affine subspaces. Distances between subspaces are then obtained by exploiting the non-Euclidean geometry of Grassmann manifolds. Experiments on challenging videos (containing object occlusions, deformations, as well as variations in pose and illumination) indicate that the proposed method achieves higher tracking accuracy than several recent discriminative trackers.

Vikas Reddy, Conrad Sanderson, Andres Sanin et al.

Robust foreground object segmentation via background modelling is a difficult problem in cluttered environments, where obtaining a clear view of the background to model is almost impossible. In this paper, we propose a method capable of robustly estimating the background and detecting regions of interest in such environments. In particular, we propose to extend the background initialisation component of a recent patch-based foreground detection algorithm with an elaborate technique based on Markov Random Fields, where the optimal labelling solution is computed using iterated conditional modes. Rather than relying purely on local temporal statistics, the proposed technique takes into account the spatial continuity of the entire background. Experiments with several tracking algorithms on the CAVIAR dataset indicate that the proposed method leads to considerable improvements in object tracking accuracy, when compared to methods based on Gaussian mixture models and feature histograms.

Masoud Faraki, Maziar Palhang, Conrad Sanderson

Representing videos by densely extracted local space-time features has recently become a popular approach for analysing actions. In this paper, we tackle the problem of categorising human actions by devising Bag of Words (BoW) models based on covariance matrices of spatio-temporal features, with the features formed from histograms of optical flow. Since covariance matrices form a special type of Riemannian manifold, the space of Symmetric Positive Definite (SPD) matrices, non-Euclidean geometry should be taken into account while discriminating between covariance matrices. To this end, we propose to embed SPD manifolds to Euclidean spaces via a diffeomorphism and extend the BoW approach to its Riemannian version. The proposed BoW approach takes into account the manifold geometry of SPD matrices during the generation of the codebook and histograms. Experiments on challenging human action datasets show that the proposed method obtains notable improvements in discrimination accuracy, in comparison to several state-of-the-art methods.

Arnold Wiliem, Conrad Sanderson, Yongkang Wong et al.

This paper describes a novel system for automatic classification of images obtained from Anti-Nuclear Antibody (ANA) pathology tests on Human Epithelial type 2 (HEp-2) cells using the Indirect Immunofluorescence (IIF) protocol. The IIF protocol on HEp-2 cells has been the hallmark method to identify the presence of ANAs, due to its high sensitivity and the large range of antigens that can be detected. However, it suffers from numerous shortcomings, such as being subjective as well as time and labour intensive. Computer Aided Diagnostic (CAD) systems have been developed to address these problems, which automatically classify a HEp-2 cell image into one of its known patterns (eg. speckled, homogeneous). Most of the existing CAD systems use handpicked features to represent a HEp-2 cell image, which may only work in limited scenarios. We propose a novel automatic cell image classification method termed Cell Pyramid Matching (CPM), which is comprised of regional histograms of visual words coupled with the Multiple Kernel Learning framework. We present a study of several variations of generating histograms and show the efficacy of the system on two publicly available datasets: the ICPR HEp-2 cell classification contest dataset and the SNPHEp-2 dataset.

Andres Sanin, Conrad Sanderson, Mehrtash T. Harandi et al.

For covariance-based image descriptors, taking into account the curvature of the corresponding feature space has been shown to improve discrimination performance. This is often done through representing the descriptors as points on Riemannian manifolds, with the discrimination accomplished on a tangent space. However, such treatment is restrictive as distances between arbitrary points on the tangent space do not represent true geodesic distances, and hence do not represent the manifold structure accurately. In this paper we propose a general discriminative model based on the combination of several tangent spaces, in order to preserve more details of the structure. The model can be used as a weak learner in a boosting-based pedestrian detection framework. Experiments on the challenging INRIA and DaimlerChrysler datasets show that the proposed model leads to considerably higher performance than methods based on histograms of oriented gradients as well as previous Riemannian-based techniques.

Azadeh Alavi, Arnold Wiliem, Kun Zhao et al.

Recent advances suggest that encoding images through Symmetric Positive Definite (SPD) matrices and then interpreting such matrices as points on Riemannian manifolds can lead to increased classification performance. Taking into account manifold geometry is typically done via (1) embedding the manifolds in tangent spaces, or (2) embedding into Reproducing Kernel Hilbert Spaces (RKHS). While embedding into tangent spaces allows the use of existing Euclidean-based learning algorithms, manifold shape is only approximated which can cause loss of discriminatory information. The RKHS approach retains more of the manifold structure, but may require non-trivial effort to kernelise Euclidean-based learning algorithms. In contrast to the above approaches, in this paper we offer a novel solution that allows SPD matrices to be used with unmodified Euclidean-based learning algorithms, with the true manifold shape well-preserved. Specifically, we propose to project SPD matrices using a set of random projection hyperplanes over RKHS into a random projection space, which leads to representing each matrix as a vector of projection coefficients. Experiments on face recognition, person re-identification and texture classification show that the proposed approach outperforms several recent methods, such as Tensor Sparse Coding, Histogram Plus Epitome, Riemannian Locality Preserving Projection and Relational Divergence Classification.

Azadeh Alavi, Yan Yang, Mehrtash Harandi et al.

Person re-identification is particularly challenging due to significant appearance changes across separate camera views. In order to re-identify people, a representative human signature should effectively handle differences in illumination, pose and camera parameters. While general appearance-based methods are modelled in Euclidean spaces, it has been argued that some applications in image and video analysis are better modelled via non-Euclidean manifold geometry. To this end, recent approaches represent images as covariance matrices, and interpret such matrices as points on Riemannian manifolds. As direct classification on such manifolds can be difficult, in this paper we propose to represent each manifold point as a vector of similarities to class representers, via a recently introduced form of Bregman matrix divergence known as the Stein divergence. This is followed by using a discriminative mapping of similarity vectors for final classification. The use of similarity vectors is in contrast to the traditional approach of embedding manifolds into tangent spaces, which can suffer from representing the manifold structure inaccurately. Comparative evaluations on benchmark ETHZ and iLIDS datasets for the person re-identification task show that the proposed approach obtains better performance than recent techniques such as Histogram Plus Epitome, Partial Least Squares, and Symmetry-Driven Accumulation of Local Features.

Shaokang Chen, Arnold Wiliem, Conrad Sanderson et al.

Traditional nearest points methods use all the samples in an image set to construct a single convex or affine hull model for classification. However, strong artificial features and noisy data may be generated from combinations of training samples when significant intra-class variations and/or noise occur in the image set. Existing multi-model approaches extract local models by clustering each image set individually only once, with fixed clusters used for matching with various image sets. This may not be optimal for discrimination, as undesirable environmental conditions (eg. illumination and pose variations) may result in the two closest clusters representing different characteristics of an object (eg. frontal face being compared to non-frontal face). To address the above problem, we propose a novel approach to enhance nearest points based methods by integrating affine/convex hull classification with an adapted multi-model approach. We first extract multiple local convex hulls from a query image set via maximum margin clustering to diminish the artificial variations and constrain the noise in local convex hulls. We then propose adaptive reference clustering (ARC) to constrain the clustering of each gallery image set by forcing the clusters to have resemblance to the clusters in the query image set. By applying ARC, noisy clusters in the query set can be discarded. Experiments on Honda, MoBo and ETH-80 datasets show that the proposed method outperforms single model approaches and other recent techniques, such as Sparse Approximated Nearest Points, Mutual Subspace Method and Manifold Discriminant Analysis.

Johanna Carvajal, Chris McCool, Conrad Sanderson

We present a novel approach to video summarisation that makes use of a Bag-of-visual-Textures (BoT) approach. Two systems are proposed, one based solely on the BoT approach and another which exploits both colour information and BoT features. On 50 short-term videos from the Open Video Project we show that our BoT and fusion systems both achieve state-of-the-art performance, obtaining an average F-measure of 0.83 and 0.86 respectively, a relative improvement of 9% and 13% when compared to the previous state-of-the-art. When applied to a new underwater surveillance dataset containing 33 long-term videos, the proposed system reduces the amount of footage by a factor of 27, with only minor degradation in the information content. This order of magnitude reduction in video data represents significant savings in terms of time and potential labour cost when manually reviewing such footage.

Sareh Shirazi, Mehrtash T. Harandi, Brian C. Lovell et al.

A robust visual tracking system requires an object appearance model that is able to handle occlusion, pose, and illumination variations in the video stream. This can be difficult to accomplish when the model is trained using only a single image. In this paper, we first propose a tracking approach based on affine subspaces (constructed from several images) which are able to accommodate the abovementioned variations. We use affine subspaces not only to represent the object, but also the candidate areas that the object may occupy. We furthermore propose a novel approach to measure affine subspace-to-subspace distance via the use of non-Euclidean geometry of Grassmann manifolds. The tracking problem is then considered as an inference task in a Markov Chain Monte Carlo framework via particle filtering. Quantitative evaluation on challenging video sequences indicates that the proposed approach obtains considerably better performance than several recent state-of-the-art methods such as Tracking-Learning-Detection and MILtrack.

Mehrtash Harandi, Richard Hartley, Chunhua Shen et al.

Sparsity-based representations have recently led to notable results in various visual recognition tasks. In a separate line of research, Riemannian manifolds have been shown useful for dealing with features and models that do not lie in Euclidean spaces. With the aim of building a bridge between the two realms, we address the problem of sparse coding and dictionary learning over the space of linear subspaces, which form Riemannian structures known as Grassmann manifolds. To this end, we propose to embed Grassmann manifolds into the space of symmetric matrices by an isometric mapping. This in turn enables us to extend two sparse coding schemes to Grassmann manifolds. Furthermore, we propose closed-form solutions for learning a Grassmann dictionary, atom by atom. Lastly, to handle non-linearity in data, we extend the proposed Grassmann sparse coding and dictionary learning algorithms through embedding into Hilbert spaces. Experiments on several classification tasks (gender recognition, gesture classification, scene analysis, face recognition, action recognition and dynamic texture classification) show that the proposed approaches achieve considerable improvements in discrimination accuracy, in comparison to state-of-the-art methods such as kernelized Affine Hull Method and graph-embedding Grassmann discriminant analysis.

Mehrtash Harandi, Conrad Sanderson, Chunhua Shen et al.

Recent advances in computer vision and machine learning suggest that a wide range of problems can be addressed more appropriately by considering non-Euclidean geometry. In this paper we explore sparse dictionary learning over the space of linear subspaces, which form Riemannian structures known as Grassmann manifolds. To this end, we propose to embed Grassmann manifolds into the space of symmetric matrices by an isometric mapping, which enables us to devise a closed-form solution for updating a Grassmann dictionary, atom by atom. Furthermore, to handle non-linearity in data, we propose a kernelised version of the dictionary learning algorithm. Experiments on several classification tasks (face recognition, action recognition, dynamic texture classification) show that the proposed approach achieves considerable improvements in discrimination accuracy, in comparison to state-of-the-art methods such as kernelised Affine Hull Method and graph-embedding Grassmann discriminant analysis.

Mehrtash T. Harandi, Conrad Sanderson, Richard Hartley et al.

Recent advances suggest that a wide range of computer vision problems can be addressed more appropriately by considering non-Euclidean geometry. This paper tackles the problem of sparse coding and dictionary learning in the space of symmetric positive definite matrices, which form a Riemannian manifold. With the aid of the recently introduced Stein kernel (related to a symmetric version of Bregman matrix divergence), we propose to perform sparse coding by embedding Riemannian manifolds into reproducing kernel Hilbert spaces. This leads to a convex and kernel version of the Lasso problem, which can be solved efficiently. We furthermore propose an algorithm for learning a Riemannian dictionary (used for sparse coding), closely tied to the Stein kernel. Experiments on several classification tasks (face recognition, texture classification, person re-identification) show that the proposed sparse coding approach achieves notable improvements in discrimination accuracy, in comparison to state-of-the-art methods such as tensor sparse coding, Riemannian locality preserving projection, and symmetry-driven accumulation of local features.

Yongkang Wong, Conrad Sanderson, Sandra Mau et al.

While existing face recognition systems based on local features are robust to issues such as misalignment, they can exhibit accuracy degradation when comparing images of differing resolutions. This is common in surveillance environments where a gallery of high resolution mugshots is compared to low resolution CCTV probe images, or where the size of a given image is not a reliable indicator of the underlying resolution (eg. poor optics). To alleviate this degradation, we propose a compensation framework which dynamically chooses the most appropriate face recognition system for a given pair of image resolutions. This framework applies a novel resolution detection method which does not rely on the size of the input images, but instead exploits the sensitivity of local features to resolution using a probabilistic multi-region histogram approach. Experiments on a resolution-modified version of the "Labeled Faces in the Wild" dataset show that the proposed resolution detector frontend obtains a 99% average accuracy in selecting the most appropriate face recognition system, resulting in higher overall face discrimination accuracy (across several resolutions) compared to the individual baseline face recognition systems.

Arnold Wiliem, Yongkang Wong, Conrad Sanderson et al.

The Anti-Nuclear Antibody (ANA) clinical pathology test is commonly used to identify the existence of various diseases. A hallmark method for identifying the presence of ANAs is the Indirect Immunofluorescence method on Human Epithelial (HEp-2) cells, due to its high sensitivity and the large range of antigens that can be detected. However, the method suffers from numerous shortcomings, such as being subjective as well as time and labour intensive. Computer Aided Diagnostic (CAD) systems have been developed to address these problems, which automatically classify a HEp-2 cell image into one of its known patterns (eg., speckled, homogeneous). Most of the existing CAD systems use handpicked features to represent a HEp-2 cell image, which may only work in limited scenarios. In this paper, we propose a cell classification system comprised of a dual-region codebook-based descriptor, combined with the Nearest Convex Hull Classifier. We evaluate the performance of several variants of the descriptor on two publicly available datasets: ICPR HEp-2 cell classification contest dataset and the new SNPHEp-2 dataset. To our knowledge, this is the first time codebook-based descriptors are applied and studied in this domain. Experiments show that the proposed system has consistent high performance and is more robust than two recent CAD systems.

Andres Sanin, Conrad Sanderson, Brian C. Lovell

This paper presents a survey and a comparative evaluation of recent techniques for moving cast shadow detection. We identify shadow removal as a critical step for improving object detection and tracking. The survey covers methods published during the last decade, and places them in a feature-based taxonomy comprised of four categories: chromacity, physical, geometry and textures. A selection of prominent methods across the categories is compared in terms of quantitative performance measures (shadow detection and discrimination rates, colour desaturation) as well as qualitative observations. Furthermore, we propose the use of tracking performance as an unbiased approach for determining the practical usefulness of shadow detection methods. The evaluation indicates that all shadow detection approaches make different contributions and all have individual strength and weaknesses. Out of the selected methods, the geometry-based technique has strict assumptions and is not generalisable to various environments, but it is a straightforward choice when the objects of interest are easy to model and their shadows have different orientation. The chromacity based method is the fastest to implement and run, but it is sensitive to noise and less effective in low saturated scenes. The physical method improves upon the accuracy of the chromacity method by adapting to local shadow models, but fails when the spectral properties of the objects are similar to that of the background. The small-region texture based method is especially robust for pixels whose neighbourhood is textured, but may take longer to implement and is the most computationally expensive. The large-region texture based method produces the most accurate results, but has a significant computational load due to its multiple processing steps.

Vikas Reddy, Conrad Sanderson, Brian C. Lovell

A robust and efficient anomaly detection technique is proposed, capable of dealing with crowded scenes where traditional tracking based approaches tend to fail. Initial foreground segmentation of the input frames confines the analysis to foreground objects and effectively ignores irrelevant background dynamics. Input frames are split into non-overlapping cells, followed by extracting features based on motion, size and texture from each cell. Each feature type is independently analysed for the presence of an anomaly. Unlike most methods, a refined estimate of object motion is achieved by computing the optical flow of only the foreground pixels. The motion and size features are modelled by an approximated version of kernel density estimation, which is computationally efficient even for large training datasets. Texture features are modelled by an adaptively grown codebook, with the number of entries in the codebook selected in an online fashion. Experiments on the recently published UCSD Anomaly Detection dataset show that the proposed method obtains considerably better results than three recent approaches: MPPCA, social force, and mixture of dynamic textures (MDT). The proposed method is also several orders of magnitude faster than MDT, the next best performing method.

Yongkang Wong, Shaokang Chen, Sandra Mau et al.

In video based face recognition, face images are typically captured over multiple frames in uncontrolled conditions, where head pose, illumination, shadowing, motion blur and focus change over the sequence. Additionally, inaccuracies in face localisation can also introduce scale and alignment variations. Using all face images, including images of poor quality, can actually degrade face recognition performance. While one solution it to use only the "best" subset of images, current face selection techniques are incapable of simultaneously handling all of the abovementioned issues. We propose an efficient patch-based face image quality assessment algorithm which quantifies the similarity of a face image to a probabilistic face model, representing an "ideal" face. Image characteristics that affect recognition are taken into account, including variations in geometric alignment (shift, rotation and scale), sharpness, head pose and cast shadows. Experiments on FERET and PIE datasets show that the proposed algorithm is able to identify images which are simultaneously the most frontal, aligned, sharp and well illuminated. Further experiments on a new video surveillance dataset (termed ChokePoint) show that the proposed method provides better face subsets than existing face selection techniques, leading to significant improvements in recognition accuracy.

Sandra Mau, Shaokang Chen, Conrad Sanderson et al.

Balancing computational efficiency with recognition accuracy is one of the major challenges in real-world video-based face recognition. A significant design decision for any such system is whether to process and use all possible faces detected over the video frames, or whether to select only a few "best" faces. This paper presents a video face recognition system based on probabilistic Multi-Region Histograms to characterise performance trade-offs in: (i) selecting a subset of faces compared to using all faces, and (ii) combining information from all faces via clustering. Three face selection metrics are evaluated for choosing a subset: face detection confidence, random subset, and sequential selection. Experiments on the recently introduced MOBIO dataset indicate that the usage of all faces through clustering always outperformed selecting only a subset of faces. The experiments also show that the face selection metric based on face detection confidence generally provides better recognition performance than random or sequential sampling. Moreover, the optimal number of faces varies drastically across selection metric and subsets of MOBIO. Given the trade-offs between computational effort, recognition accuracy and robustness, it is recommended that face feature clustering would be most advantageous in batch processing (particularly for video-based watchlists), whereas face selection methods should be limited to applications with significant computational restrictions.

Andres Sanin, Conrad Sanderson, Mehrtash T. Harandi et al.

We propose a new action and gesture recognition method based on spatio-temporal covariance descriptors and a weighted Riemannian locality preserving projection approach that takes into account the curved space formed by the descriptors. The weighted projection is then exploited during boosting to create a final multiclass classification algorithm that employs the most useful spatio-temporal regions. We also show how the descriptors can be computed quickly through the use of integral video representations. Experiments on the UCF sport, CK+ facial expression and Cambridge hand gesture datasets indicate superior performance of the proposed method compared to several recent state-of-the-art techniques. The proposed method is robust and does not require additional processing of the videos, such as foreground detection, interest-point detection or tracking.