Fast and successful searching for an object in a multimedia database is a highly desirable functionality. Several approaches to content based retrieval for multimedia databases can be found in the literature [9,10,12,14,17]. The approach we consider is feature extraction. A feature can be seen as a way to present simple information like the texture, color and spatial information of an image, or the pitch, frequency of a sound etc. In this paper we present a method for feature extraction on texture and spatial similarity, using fractal coding techniques. Our method is based upon the observation that the coefficients describing the fractal code of an image, contain very useful information about the structural content of the image. We apply simple statistics on information produced by fractal image coding. The statistics reveal features and require a small amount of storage. Several invariances are a consequence of the used methods: size, global contrast, orientation.
DOCUMENT
Sensor systems can be deployed in the homes of older adults living alone for functional health assessments. Their information is very useful for health care specialists. The problem lies in developing person independent models while facing a large variability in behavior. We address this problem by, first, proposing a new feature extraction method for data from ambient motion sensors. The method uses functional similarities between houses and daily structure to extract meaningful features. Second, we propose a change-based approach for analyzing data, taking difference scores of both the sensor features and health metrics. To evaluate our approach, experiments on longitudinal data were conducted, where the relationship between sensor data and health measurements was modeled with linear regression and (nonlinear) regression forests. These experiments show that the change-based approach yields better results and that the resulting models can be used as a reliable metric for (functional) health. In addition, feature analysis can help health care specialists understand relevant aspects of behavior. Prediction of health metrics is possible even with simple sensors. With such sensors, it is possible to detect problems and health decline in an early stage. This will have great impact on clinical practice.
DOCUMENT
The huge number of images shared on the Web makes effective cataloguing methods for efficient storage and retrieval procedures specifically tailored on the end-user needs a very demanding and crucial issue. In this paper, we investigate the applicability of Automatic Image Annotation (AIA) for image tagging with a focus on the needs of database expansion for a news broadcasting company. First, we determine the feasibility of using AIA in such a context with the aim of minimizing an extensive retraining whenever a new tag needs to be incorporated in the tag set population. Then, an image annotation tool integrating a Convolutional Neural Network model (AlexNet) for feature extraction and a K-Nearest-Neighbours classifier for tag assignment to images is introduced and tested. The obtained performances are very promising addressing the proposed approach as valuable to tackle the problem of image tagging in the framework of a broadcasting company, whilst not yet optimal for integration in the business process.
DOCUMENT
The security of online assessments is a major concern due to widespread cheating. One common form of cheating is impersonation, where students invite unauthorized persons to take assessments on their behalf. Several techniques exist to handle impersonation. Some researchers recommend use of integrity policy, but communicating the policy effectively to the students is a challenge. Others propose authentication methods like, password and fingerprint; they offer initial authentication but are vulnerable thereafter. Face recognition offers post-login authentication but necessitates additional hardware. Keystroke Dynamics (KD) has been used to provide post-login authentication without any additional hardware, but its use is limited to subjective assessment. In this work, we address impersonation in assessments with Multiple Choice Questions (MCQ). Our approach combines two key strategies: reinforcement of integrity policy for prevention, and keystroke-based random authentication for detection of impersonation. To the best of our knowledge, it is the first attempt to use keystroke dynamics for post-login authentication in the context of MCQ. We improve an online quiz tool for the data collection suited to our needs and use feature engineering to address the challenge of high-dimensional keystroke datasets. Using machine learning classifiers, we identify the best-performing model for authenticating the students. The results indicate that the highest accuracy (83%) is achieved by the Isolation Forest classifier. Furthermore, to validate the results, the approach is applied to Carnegie Mellon University (CMU) benchmark dataset, thereby achieving an improved accuracy of 94%. Though we also used mouse dynamics for authentication, but its subpar performance leads us to not consider it for our approach.
DOCUMENT
The performance of human-robot collaboration tasks can be improved by incorporating predictions of the human collaborator's movement intentions. These predictions allow a collaborative robot to both provide appropriate assistance and plan its own motion so it does not interfere with the human. In the specific case of human reach intent prediction, prior work has divided the task into two pieces: recognition of human activities and prediction of reach intent. In this work, we propose a joint model for simultaneous recognition of human activities and prediction of reach intent based on skeletal pose. Since future reach intent is tightly linked to the action a person is performing at present, we hypothesize that this joint model will produce better performance on the recognition and prediction tasks than past approaches. In addition, our approach incorporates a simple human kinematic model which allows us to generate features that compactly capture the reachability of objects in the environment and the motion cost to reach those objects, which we anticipate will improve performance. Experiments using the CAD-120 benchmark dataset show that both the joint modeling approach and the human kinematic features give improved F1 scores versus the previous state of the art.
DOCUMENT
In a recent official statement, Google highlighted the negative effects of fake reviews on review websites and specifically requested companies not to buy and users not to accept payments to provide fake reviews (Google, 2019). Also, governmental authorities started acting against organisations that show to have a high number of fake reviews on their apps (DigitalTrends, 2018; Gov UK, 2020; ACM, 2017). However, while the phenomenon of fake reviews is well-known in industries as online journalism and business and travel portals, it remains a difficult challenge in software engineering (Martens & Maalej, 2019). Fake reviews threaten the reputation of an organisation and lead to a disvalued source to determine the public opinion about brands. Negative fake reviews can lead to confusion for customers and a loss of sales. Positive fake reviews might also lead to wrong insights about real users’ needs and requirements. Although fake reviews have been studied for a while now, there are only a limited number of spam detection models available for companies to protect their corporate reputation. Especially in times with the coronavirus, organisations need to put extra focus on online presence and limit the amount of negative input that affects their competitive position which can even lead to business loss. Given state-of-the-art derived features that can be engineered from review texts, a spam detector based on supervised machine learning is derived in an experiment that performs quite well on the well-known Amazon Mechanical Turk dataset.
MULTIFILE
Adopted on the fifteenth anniversary of resolution 1325, Security Council resolution 2242 has recognized for the first time the substantial link between climate change and the “Women, Peace and Security” (WPS) framework. Despite this landmark resolution, the intersections of environmental factors, conflict and violence against women remain largely absent from the Security Council's WPS agenda. Competition over natural resources is generally understood as a driver of conflict. The risk of insecurity and conflict are further increased by environmental degradation and climate change. It is therefore clear that the environment and natural resources must be integrated into the WPS agenda. This should necessarily include a discussion of indigenous rights to land and the gender-related dimensions of environmental factors. Indigenous women are disproportionately affected by environmental degradation, caused by resource extraction and increasingly compounded by climatic changes. This in turn exacerbates other vulnerabilities, including sexual and gender-based violence and other forms of marginalization. This article argues, by reference to the situation in West Papua, that unfettered resource extraction not only amplifies vulnerabilities and exacerbates preexisting inequalities stemming from colonial times, it also gives rise to gendered consequences flowing from the damage wreaked on the natural environment and thus poses a danger to international peace and security. As such, the Security Council's failure to recognize the continuous struggle of women in indigenous and rural communities against extractive economies and climate change impact as a security risk forms a serious lacuna within its WPS agenda. Originally published by Oxford University Press in Global Studies Quarterly, Volume 1, Issue 3, September 2021, ksab018, https://doi.org/10.1093/isagsq/ksab018
MULTIFILE
Within recent years, Financial Credit Risk Assessment (FCRA) has become an increasingly important issue within the financial industry. Therefore, the search for features that can predict the credit risk of an organization has increased. Using multiple statistical techniques, a variance of features has been proposed. Applying a structured literature review, 258 papers have been selected. From the selected papers, 835 features have been identified. The features have been analyzed with respect to the type of feature, the information sources needed and the type of organization that applies the features. Based on the results of the analysis, the features have been plotted in the FCRA Model. The results show that most features focus on hard information from a transactional source, based on official information with a high latency. In this paper, we readdress and -present our earlier work [1]. We extended the previous research with more detailed descriptions of the related literature, findings, and results, which provides a grounded basis from which further research on FCRA can be conducted.
DOCUMENT
Ambient activity monitoring systems produce large amounts of data, which can be used for health monitoring. The problem is that patterns in this data reflecting health status are not identified yet. In this paper the possibility is explored of predicting the functional health status (the motor score of AMPS = Assessment of Motor and Process Skills) of a person from data of binary ambient sensors. Data is collected of five independently living elderly people. Based on expert knowledge, features are extracted from the sensor data and several subsets are selected. We use standard linear regression and Gaussian processes for mapping the features to the functional status and predict the status of a test person using a leave-oneperson-out cross validation. The results show that Gaussian processes perform better than the linear regression model, and that both models perform better with the basic feature set than with location or transition based features. Some suggestions are provided for better feature extraction and selection for the purpose of health monitoring. These results indicate that automated functional health assessment is possible, but some challenges lie ahead. The most important challenge is eliciting expert knowledge and translating that into quantifiable features.
DOCUMENT
Industrial robot manipulators are widely used for repetitive applications that require high precision, like pick-and-place. In many cases, the movements of industrial robot manipulators are hard-coded or manually defined, and need to be adjusted if the objects being manipulated change position. To increase flexibility, an industrial robot should be able to adjust its configuration in order to grasp objects in variable/unknown positions. This can be achieved by off-the-shelf vision-based solutions, but most require prior knowledge about each object tobe manipulated. To address this issue, this work presents a ROS-based deep reinforcement learning solution to robotic grasping for a Collaborative Robot (Cobot) using a depth camera. The solution uses deep Q-learning to process the color and depth images and generate a greedy policy used to define the robot action. The Q-values are estimated using Convolutional Neural Network (CNN) based on pre-trained models for feature extraction. Experiments were carried out in a simulated environment to compare the performance of four different pre-trained CNNmodels (RexNext, MobileNet, MNASNet and DenseNet). Results showthat the best performance in our application was reached by MobileNet,with an average of 84 % accuracy after training in simulated environment.
DOCUMENT
