Search

Retinal and optic disc images are used to assess changes in the retinal vasculature. These can be changes associated with diseases such as diabetic retinopathy and glaucoma or induced using ophthalmodynamometry to measure arterial and venous pressure. Key steps toward automating the assessment of these changes are the segmentation and classification of the veins and arteries. However, such segmentation and classification are still required to be manually labelled by experts. Such automated labelling is challenging because of the complex morphology, anatomical variations, alterations due to disease and scarcity of labelled data for algorithm development. We present a deep machine learning solution called the multiscale guided attention network for retinal artery and vein segmentation and classification (MSGANet-RAV).MSGANet-RAV was developed and tested on 383 colour clinical optic disc images from LEI-CENTRAL, constructed in-house and 40 colour fundus images from the AV-DRIVE public dataset. The datasets have a mean optic disc occupancy per image of 60.6% and 2.18%, respectively. MSGANet-RAV is a U-shaped encoder-decoder network, where the encoder extracts multiscale features, and the decoder includes a sequence of self-attention modules. The self-attention modules explore, guide and incorporate vessel-specific structural and contextual feature information to segment and classify central optic disc and retinal vessel pixels.MSGANet-RAV achieved a pixel classification accuracy of 93.15%, sensitivity of 92.19%, and specificity of 94.13% on LEI-CENTRAL, outperforming several reference models. It similarly performed highly on AV-DRIVE with an accuracy, sensitivity and specificity of 95.48%, 93.59% and 97.27%, respectively.The results show the efficacy of MSGANet-RAV for identifying central optic disc and retinal arteries and veins. The method can be used in automated systems designed to assess vascular changes in retinal and optic disc images quantitatively.

Deep learning models have been demonstrated vulnerable to adversarial attacks even with imperceptible perturbations. As such, the reliability of existing deep neural networks-based autonomous driving systems can suffer. However, deep 3D models have applications in various Cyber-Physical Systems (CPSs) with safety-critical requirements, particularly autonomous driving systems. In this paper, the robustness of deep 3D object detection models under adversarial point cloud perturbations has been investigated. A novel method is developed to generate 3D adversarial examples from point cloud perturbations, which are common due to the intrinsic characteristics of the data captured by 3D sensors, e.g., LiDAR. The generation of adversarial samples is supervised by a dual loss, which constitutes an adversarial loss and a perturbation loss. The adversarial loss produces a point cloud with the property of aggressiveness, while the perturbation loss enforces the produced point cloud subject to visual imperception. We demonstrate that the method can successfully attack 3D object detection models in most cases, and expose their vulnerability to physical-world attacks in the form of point cloud perturbations. We perform a thorough evaluation of popular deep 3D object detectors in an adversarial setting on the KITTI vision benchmark. Experimental results show that current deep 3D object detection models are susceptible to adversarial attacks in the context of autonomous driving, and their performances are degraded by a large margin in the presence of adversarial point clouds generated by the proposed method.

This paper proposes a novel image set classification technique based on the concept of linear regression. Unlike most other approaches, the proposed technique does not require any training. We represent the gallery image sets as subspaces in a high dimensional space. Class specific gallery subspaces are used to estimate regression models for each image in the test image set. Images of the test set are then projected onto the gallery subspaces. The residuals, calculated using the Euclidean distance between the original and the projected test images, are used as the distance metric. Three different strategies are devised to decide on the final class of the test image set. We extensively evaluated the proposed technique using both low resolution and noisy images and with less gallery data to assess the suitability of our technique for the tasks of surveillance and video-based face recognition. The experiments show that the proposed technique achieves superior classification accuracy and has a faster execution time compared with existing techniques, especially under the challenging conditions of low resolution and a limited amount of gallery and test data.

We consider the 3D object recognition problem from the perspective of the lack of labelled data. In this paper, we propose a novel progressive conditional generative adversarial network (PC-GAN) for 3D object recognition by conditioning the input with progressive learning strategies. PC-GAN is a powerful adversarial model whose generator automatically produces realistic 3D objects with annotations, and the discriminator distinguishes them from the training distribution and recognizes their categories. We train the discriminative classifier simultaneously with the generator to predict the class label by embedding a SoftMax classifier. Progressive learning uses input samples from lower to higher resolutions to increase the generator performance gradually and produce informative objects for a certain class of objects. The key idea of adopting progressing learning is to mitigate overshoots issues of the discriminator and increase variations in the generated objects by learning progressively. This strategy helps the generator to produce more realistic synthetic objects and improve the active classification performance of the discriminator. Our proposed PC-GAN is trained for object classification in a supervised manner and the performance is evaluated on two public datasets. Experimental results demonstrate that our adversarial PC-GAN outperforms the existing volumetric discriminative classifiers in term of classification accuracy.

Missing data is a common problem faced with real-world datasets. Imputation is a widely used technique to estimate the missing data. State-of-the-art imputation approaches model the distribution of observed data to approximate the missing values. Such an approach usually models a single distribution for the entire dataset, which overlooks the class-specific characteristics of the data. Class-specific characteristics are especially useful when there is a class imbalance. We propose a new method for imputing missing data based on its class-specific characteristics by adapting the popular Conditional Generative Adversarial Networks (CGAN). Our Conditional Generative Adversarial Imputation Network (CGAIN) imputes the missing data using class-specific distributions, which can produce the best estimates for the missing values. We tested our approach on baseline datasets and achieved superior performance compared with the state-of-the-art and popular imputation approaches.

Weakly supervised semantic segmentation has been attracting increasing attention as it can alleviate the need for expensive pixel-level annotations through the use of image-level labels. Relevant methods mainly rely on the implicit object localization ability of convolutional neural networks (CNNs). However, generated object attention maps remain mostly small and incomplete. In this paper, we propose an Atrous Convolutional Feature Network (ACFN) to generate dense object attention maps. This is achieved by enhancing the context representation of image classification CNNs. More specifically, cascaded atrous convolutions are used in the middle layers to retain sufficient spatial details, and pyramidal atrous convolutions are used in the last convolutional layers to provide multi-scale context information for the extraction of object attention maps. Moreover, we propose an attentive fusion strategy to adaptively fuse the multi-scale features. Our method shows improvements over existing methods on both the PASCAL VOC 2012 and MS COCO datasets, achieving state-of-the-art performance.

Australia has a reputation for producing a reliable supply of high-quality barley in a contaminant-free climate. As a result, Australian barley is highly sought after by malting, brewing, distilling, and feed industries worldwide. Barley is traded as a variety-specific commodity on the international market for food, brewing and distilling end-use, as the intrinsic quality of the variety determines its market value. Manual identification of barley varieties by the naked eye is challenging and time-consuming for all stakeholders, including growers, grain handlers and traders. Current industrial methods for identifying barley varieties include molecular protein weights or DNA based technology, which are not only time-consuming and costly but need specific laboratory equipment. On grain receival, there is a need for efficient and low-cost solutions for barley classification to ensure accurate and effective variety segregation. This paper proposes an efficient deep learning-based technique that can classify barley varieties from RGB images. Our proposed technique takes only four milliseconds to classify an RGB image. The proposed technique outperforms the baseline method and achieves a barley classification accuracy of 94% across 14 commercial barley varieties (some highly genetically related).

Event logs are one of the most important sources of digital evidence for forensic investigation because they record essential activities on the system. In this paper, we present a comprehensive literature survey of the forensic analysis on operating system logs. We present a taxonomy of various techniques used in this area. Additionally, we discuss the tools that support the examination of the event logs. This survey also gives a review of the publicly available datasets that are used in operating system log forensics research. Finally, we suggest potential future directions on the topic of operating system log forensics.

A common approach to tackle 3D object recognition tasks is to project 3D data to multiple 2D images. Projection only captures the outline of the object, and discards the internal information that may be crucial for the recognition. In this paper, we stay in 3D and concentrate on tapping the potential of 3D representations. We present NormalNet, a voxel-based convolutional neural network (CNN) designed for 3D object recognition. The network uses normal vectors of the object surfaces as input, which demonstrate stronger discrimination capability than binary voxels. We propose a reflection–convolution–concatenation (RCC) module to realize the conv layers, which extracts distinguishable features for 3D vision tasks while reducing the number of parameters significantly. We further improve the performance of NormalNet by combining two networks, which take normal vectors and voxels as input respectively. We carry out a series of experiments that validate the design of the network and achieve competitive performance in 3D object classification and retrieval tasks.

The variety of event categories and event boundary information have resulted in limited success for acoustic event detection systems. To deal with this, we propose to utilize the long contextual information, low-dimensional discriminant global bottleneck features and category-specific bottleneck features. By concatenating several adjacent frames together, the use of contextual information makes it easier to cope with acoustic signals with long duration. Global and category-specific bottleneck features can extract the prior knowledge of the event category and boundary, which is ideally matched by the task of an event detection system. Evaluations on the UPC-TALP and ITC-IRST databases of highly variable acoustic events demonstrate the effectiveness of the proposed approaches by achieving a 5.30% and 4.44% absolute error rate improvement respectively compared to the state of art technique.

By introducing sign constraints on the weights, this paper proposes sign constrained rectifier networks (SCRNs), whose training can be solved efficiently by the well known majorization-minimization (MM) algorithms. We prove that the proposed two-hidden-layer SCRNs, which exhibit negative weights in the second hidden layer and negative weights in the output layer, are capable of separating any two (or more) disjoint pattern sets. Furthermore, the proposed two-hidden-layer SCRNs can decompose the patterns of each class into several clusters so that each cluster is convexly separable from all the patterns from the other classes. This provides a means to learn the pattern structures and analyse the discriminant factors between different classes of patterns., Comment: 11 pages

An investigator needs to analyze a forensic timeline after a cybersecurity incident has occurred. Log entries from various sources are used to generate a forensic timeline. Finding the anomalous activities recorded in these log records is a difficult task if manual inspection or keyword searches are used. In this work, we propose a method for identifying anomalies in a forensic timeline. We use deep autoencoders as a machine learning technique to establish a baseline for normal activities in log files. Furthermore, we set an anomaly threshold of reconstruction value based on the constructed baseline. We then plot these anomalous events on a forensic timeline. Our experiments indicate that the proposed method achieves superior performance compared to other log anomaly detection methods with overall mean F1 score and accuracy of 94.036% and 96.720%, respectively.

Salient region detection without prior knowledge is a challenging task, especially for 3D deformable shapes. This paper presents a novel framework that relies on clustering of a data set derived from the scale space of the auto diffusion function. It consists of three major techniques: scalar field construction, shape segmentation initialization and salient region detection. We define the scalar field using the auto diffusion function at consecutive time scales to reveal shape features. Initial segmentation of a shape is obtained using persistence-based clustering, which is performed on the scalar field at a large time scale to capture the global shape structure. We propose two measures to assess the clustering both on a global and local level using persistent homology. From these measures, salient regions are detected during the evolution of the scalar field. Experimental results on three popular datasets demonstrate the superior performance of the proposed framework in region detection.

This paper presents a learned feature based framework for both outdoor and indoor scene labeling. This framework is combined with a discriminative feature learning process to produce the posteriors of every pixel and a novel strategy to improve the global label consistency of a scene. First, we use Convolutional Neural Networks (ConvNets) to learn the most relevant features of a scene at the multi-scale superpixel level. The effect of both trained and general ConvNets features for our scene labeling framework are investigated. Then, based on the predicted posteriors from the learned features, we propose an algorithm called Region Consistency Activation (RCA) to iteratively improve the global label consistency at different levels of the Ultrametric Contour Map (UCM). In addition, we propose a strategy to make the hyper-parameters of RCA adaptive to the test images, which results in a better generalization ability compared with the hyper-parameters tuning based RCA. Our scene labeling framework were rigorously tested on three popular scene labeling datasets: Stanford Background, SIFT Flow and NYU-Depth V2. Experiments show that our proposed method consistently produces better accuracy and visual consistency compared with the state-of-the-art methods for both outdoor and indoor scenes.

Attacks on operating system access control have become a significant and increasingly common problem. This type of security threat is recorded in a forensic artifact such as an authentication log. Forensic investigators will generally examine the log to analyze such incidents. An anomaly is highly correlated to an attacker's attempts to compromise the system. In this paper, we propose a novel method to automatically detect an anomaly in the access control log of an operating system. The logs will be first preprocessed and then clustered using an improved MajorClust algorithm to get a better cluster. This technique provides parameter-free clustering so that it automatically can produce an analysis report for the forensic investigators. The clustering results will be checked for anomalies based on a score that considers some factors such as the total members in a cluster, the frequency of the events in the log file, and the inter-arrival time of a specific activity. We also provide a graph-based visualization of logs to assist the investigators with easy analysis. Experimental results compiled on an open dataset of a Linux authentication log show that the proposed method achieved the accuracy of 83.14% in the authentication log dataset.

Background Risk scores such as CHA2DS2-VASc and HAS-BLED are used to assess stroke and bleeding risk respectively and choose appropriate antithrombotic therapy in patients with atrial fibrillation (AF). The application of ML models may improve risk prediction and identification of potential risk factors. Objective To investigate the usefulness of ML methods in estimating one-year risk of ischemic stroke and major bleeding in patients after hospitalisation with AF. Methods We identified adults with a history of non-valvular AF or atrial flutter who were admitted to a tertiary or secondary hospital in Perth, Western Australia from 2009 to 2016 using linked clinical and administrative data. Based on all the available risk factors in the data including individual risk factors in the scores, we built ML models and compared their predictive performance [Area under the receiver operating characteristic curve (AUC)] with the standard risk scores. Results There were 9,634 patients in the study cohort with a mean age of 77 years and 46% were female. 2407 patients died (n=1636) or were readmitted for ischemic stroke (n=157) and major bleeding (n=614) within one year after the first admission. All-cause death was treated as a competing risk. Gradient Boosting Machine identified nonconventional risk factors and achieved the best prediction (ischemic stroke: AUC 0.67 vs 0.64 for CHA2DS2-VASc; major bleeding: AUC 0.66 vs 0.53 for HAS-BLED). Conclusion ML models can identify nonconventional risk factors and also outperform commonly used risk scores for predicting ischemic stroke and major bleeding in patients with AF.

The rapid advances in Deep Learning (DL) techniques have enabled rapid detection, localisation, and recognition of objects from images or videos. DL techniques are now being used in many applications related to agriculture and farming. Automatic detection and classification of weeds can play an important role in weed management and so contribute to higher yields. Weed detection in crops from imagery is inherently a challenging problem because both weeds and crops have similar colours (‘green-on-green’), and their shapes and texture can be very similar at the growth phase. Also, a crop in one setting can be considered a weed in another. In addition to their detection, the recognition of specific weed species is essential so that targeted controlling mechanisms (e.g. appropriate herbicides and correct doses) can be applied. In this paper, we review existing deep learning-based weed detection and classification techniques. We cover the detailed literature on four main procedures, i.e., data acquisition, dataset preparation, DL techniques employed for detection, location and classification of weeds in crops, and evaluation metrics approaches. We found that most studies applied supervised learning techniques, they achieved high classification accuracy by fine-tuning pre-trained models on any plant dataset, and past experiments have already achieved high accuracy when a large amount of labelled data is available.

This paper presents a local feature based shape matching algorithm for expression-invariant 3D face recognition. Each 3D face is first automatically detected from a raw 3D data and normalized to achieve pose invariance. The 3D face is then represented by a set of keypoints and their associated local feature descriptors to achieve robustness to expression variations. During face recognition, a probe face is compared against each gallery face using both local feature matching and 3D point cloud registration. The number of feature matches, the average distance of matched features, and the number of closest point pairs after registration are used to measure the similarity between two 3D faces. These similarity metrics are then fused to obtain the final results. The proposed algorithm has been tested on the FRGC v2 benchmark and a high recognition performance has been achieved. It obtained the state-of-the-art results by achieving an overall rank-1 identification rate of 97.0% and an average verification rate of 99.01% at 0.001 false acceptance rate for all faces with neutral and non-neutral expressions. Further, the robustness of our algorithm under different occlusions has been demonstrated on the Bosphorus dataset.

The aim of this paper is to classify conductive material corrosion by eddy current pulsed thermography. Thermal transient images generate a large of amount of data which is difficult for accurate detection and classification of the different corrosion materials, especially with the hidden corrosion. We apply deep Boltzmann machines (DBM) network to automatically extract and classify features from the whole measured area. Corrosion classification is tested with several different machine learning based algorithms including: clustering, PCA, multi-layer DBM classifier. The performance of the proposed framework is measured in terms of accuracy, sensitivity, specificity and precision. Several experiments are performed on a dataset of eddy current signal samples for four different corrosion degrees. The results show that our method outperforms the existing algorithms in classification accuracy (97.9%), sensitivity (96.1%), precision (97.1%), and especially specificity (98.4%).

Oceanographers rely on advanced digital imaging systems to assess the health of marine ecosystems. The majority of the imagery collected by these systems do not get annotated due to lack of resources. Consequently, the expert labeled data is not enough to train dedicated deep networks. Meanwhile, in the deep learning community, much focus is on how to use pre-trained deep networks to classify out-of-domain images and transfer learning. In this paper, we leverage these advances to evaluate how well features extracted from deep neural networks transfer to underwater image classification. We propose new image features (called ResFeats) extracted from the different convolutional layers of a deep residual network pre-trained on ImageNet. We further combine the ResFeats extracted from different layers to obtain compact and powerful deep features. Moreover, we show that ResFeats consistently perform better than their CNN counterparts. Experimental results are provided to show the effectiveness of ResFeats with state-of-the-art classification accuracies on MLC, Benthoz15, EILAT and RSMAS datasets.

This paper presents a highly distinctive local surface feature called the TriSI feature for recognizing 3D objects in the presence of clutter and occlusion. For a feature point, we first construct a unique and repeatable Local Reference Frame (LRF) using the implicit geometrical information of neighboring triangular faces. We then generate three signatures from the three orthogonal coordinate axes of the LRF. These signatures are concatenated and then compressed into a TriSI feature. Finally, we propose an effective 3D object recognition algorithm based on hierarchical feature matching. We tested our TriSI feature on two popular datasets. Rigorous experimental results show that the TriSI feature was highly descriptive and outperformed existing algorithms under all levels of Gaussian noise, Laplacian noise, shot noise, varying mesh resolutions, occlusion, and clutter. Moreover, we tested our TriSI-based 3D object recognition algorithm on four standard datasets. The experimental results show that our algorithm achieved the best overall recognition results on these datasets.

We propose a novel confidence ratio (C-ratio) to be used in score-level fusion.We propose a confidence factor to be used in rank-level fusion.Confidence-based late fusion improves the identification accuracy.Confidence-based late fusion performs better than the state-of-the-art. This paper presents a confidence-based late fusion framework and its application to audio-visual biometric identification. We assign each biometric matcher a confidence value calculated from the matching scores it produces. Then a transformation of the matching scores is performed using a novel confidence-ratio (C-ratio) i.e., the ratio of a matcher confidence obtained at the test phase to the corresponding matcher confidence obtained at the training phase. We also propose modifications to the highest rank and Borda count rank fusion rules to incorporate the matcher confidence. We demonstrate by experiments that our proposed confidence-based fusion framework is more robust compared to the state-of-the-art late (score- and rank-level) fusion approaches.

Point clouds provide rich geometric information about a shape and a deep neural network can be used to learn effective and robust features. In this paper, we propose a novel local feature descriptor, which employs a Siamese network to directly learn robust features from the point clouds. We use a data representation based on the Mercator projection, then we use a Siamese network to map this projection into a 32-dimensional local descriptor. To validate the proposed method, we have compared it with seven state-of-the-art descriptor methods. Experimental results show the superiority of the proposed method compared to existing methods in terms of descriptiveness and robustness against noise and varying mesh resolutions.

Background: Predicting readmissions or mortality following hospital discharge in patients with heart failure (HF) is a major challenge. Commonly used machine learning (ML) algorithms have reported suboptimal performances in predicting HF readmissions or mortality due to class imbalance in the outcome levels (e.g. low number of readmissions vs no readmission). Moreover, the area under the receiver operating characteristic curve (AUC), a commonly used indicator of model performance, is a crude indicator for data with class imbalances. Our main aim was to develop ML models to predict 30-day HF readmission or mortality and compare various measures of model performance. Methods: We identified all Western Australian patients 65 years or older admitted for HF in 2003–08 in the linked Hospital Morbidity Data Collection. In addition to common ML models to predict 30-day HF readmissions or mortality, we also developed multi-layer perceptron (MLP) with a synthetic minority oversampling technique, considered superior in data with class imbalances. We calculated the AUC, sensitivity, and specificity of the models. Results: Of the 10,735 patients with HF, 23.6% were readmitted or died within 30 days. We observed AUCs of 0.51, 0.50, and 0.51 for linear regression, least absolute shrinkage and selection operator, and random forest models, respectively. The MLP model had a higher AUC (0.58), with a sensitivity of 57% and a specificity of 56%. Conclusion: The MLP method may be a better ML technique for data with outcome class imbalances. In addition to AUC, other metrics such as sensitivity and specificity should also be considered for such data.

Bezier curves (BC) are important tools in a wide range of diverse and challenging applications, from computer-aided design to generic object shape descriptors. A major constraint of the classical BC is that only global information concerning control points (CP) is considered, consequently there may be a sizeable gap between the BC and its control polygon (CtrlPoly), leading to a large distortion in shape representation. While BC variants like degree elevation, composite BC and refinement and subdivision narrow this gap, they increase the number of CP and thereby both the required bit-rate and computational complexity. In addition, while quasi-Bezier curves (QBC) close the gap without increasing the number of CP, they reduce the underlying distortion by only a fixed amount. This paper presents a novel contribution to BC theory, with the introduction of a dynamic Bezier curve (DBC) model, which embeds variable localised CP information into the inherently global Bezier framework, by strategically moving BC points towards the CtrlPoly. A shifting parameter (SP) is defined that enables curves lying within the region between the BC and CtrlPoly to be generated, with no commensurate increase in CP. DBC provides a flexible rate-distortion (RD) criterion for shape coding applications, with a theoretical model for determining the optimal SP value for any admissible distortion being formulated. Crucially DBC retains core properties of the classical BC, including the convex hull and affine invariance, and can be seamlessly integrated into both the vertex-based shape coding and shape descriptor frameworks to improve their RD performance. DBC has been empirically tested upon a number of natural and synthetically shaped objects, with qualitative and quantitative results confirming its consistently superior shape approximation performance, compared with the classical BC, QBC and other established BC-based shape descriptor techniques.

This paper addresses a fundamental limitation of the existing distortion measures embedded in vertex-based operational-rate-distortion shape coding techniques by introducing a new accurate distortion measurement algorithm based upon the actual distance, rather than either the shortest absolute distance or a distortion band.