High-level Interaction Research Articles

Evolutionary feature subsets selection based on interaction information for high dimensional imbalanced data classification

With the advent of technology in various scientific fields, high dimensional data are becoming abundant. A general approach to tackle the resulting challenges is to reduce data dimensionality through feature selection. Traditional feature selection approaches concentrate on selecting relevant features and ignoring irrelevant or redundant ones. However, most of these approaches neglect feature interactions. On the other hand, some datasets have imbalanced classes, which may result in biases towards the majority class. The main goal of this paper is to propose a novel feature selection method based on the interaction information (II) to provide higher level interaction analysis and improve the search procedure in the feature space. In this regard, an evolutionary feature subset selection algorithm based on interaction information is proposed, which consists of three stages. At the first stage, candidate features and candidate feature pairs are identified using traditional feature weighting approaches such as symmetric uncertainty (SU) and bivariate interaction information. In the second phase, candidate feature subsets are formed and evaluated using multivariate interaction information. Finally, the best candidate feature subsets are selected using dominant/dominated relationships. The proposed algorithm is compared with some other feature selection algorithms including mRMR, WJMI, IWFS, IGFS, DCSF, IWFS, K_OFSD, WFLNS, Information Gain and ReliefF in terms of the number of selected features, classification accuracy, F-measure and algorithm stability using three different classifiers, namely KNN, NB, and CART. The results justify the improvement of classification accuracy and the robustness of the proposed method in comparison with the other approaches.

Speech Act Theory as an Evaluation Tool for Human–Agent Communication

Effective communication in task-oriented situations requires high-level interactions. For human–agent collaboration, tasks need to be coordinated in a way that ensures mutual understanding. Speech Act Theory (SAT) aims to understand how utterances can be used to achieve actions. SAT consists of three components: locutionary act, illocutionary act, and perlocutionary act. This paper evaluates the agent’s verbal communication while collaborating with humans. SAT was used to anatomize the structure of the agent’s speech acts (locutionary acts), the agent’s intention behind the speech acts (illocutionary acts), and the effects on the human’s mental state (perlocutionary acts). Moreover, this paper studies the impact of human perceptions of the agent’s speech acts on the perception of collaborative performance with the agent.

New data acquisition control software developments for the simultaneous control of ten intelligent overview video cameras at W7-X

In the past two campaigns of Wendelstein 7-X (W7-X) stellarator the overview video diagnostics played a key role in the daily experiments. The current software implementations went through numerous improvements and changes according to the continuously changing requirements. However, while the control software could handle all the needs, the changes reached a point where the redesign and re-implementation became necessary to ensure high reliability. In the next campaign of W7-X real-time data streaming will be implemented to W7-X’s Archive, which can be practically a limitless real-time data backup. Other improvements involve high-level interaction with the trigger-time-event (TTE) network, intense graphical user interface (GUI) improvements such as the grouping of cameras, visual display of measurement settings (e.g. exposure and readout timing). This new software package is arranged into two standalone pieces, similarly as in the previous version. One of the main upgrades is that the software will be able to send so-called event descriptions over the network during the measurement to a central event display, where events from several diagnostics will be visualized. This feature can provide essential information for machine operators, as they can visually monitor dangerous events inside the machine in real-time. This paper will present the detailed experiences, new requirements and detailed design, implementation and testing of this new software package.

Robot-Based Training for People With Mild Cognitive Impairment

Current research suggests that physical activity has positive effects on the overall state of people with mild cognitive impairment (MCI) and that the use of new technologies could help those people to keep and improve their postural control and motor skills. In a pilot study, we investigated the use of a device based on a mobile robot for the motor activation of people with MCI, with the goal of investigating the possible influence of such an activity on the cognitive level. The device is an omnidirectional robot platform with handlebars and an integrated force-torque sensor to enable direct interaction with a user. A passive interaction controller with high-level interaction modes for training scenarios is developed. The device and the training are evaluated with ten participants (eight males and two females) from 66 to 78 years of age with MCI in five 1-h sessions. Here, we present the data gathered with the device and the evaluation of the user experience. The results for the precision of controlling the device and the time performance show that the users got better during the training week for most tasks. The participants felt safe during the training and managed to adapt to changes in the device's high-level behavior. Overall, the current results suggest that the proposed training is feasible and the device is suitable for the training. Future work will evaluate pre- and post-training MRI data and data from neuropsychological and memory testing to investigate impacts of the training with the robotic device on the cognitive level of a user.

Visbrain: A Multi-Purpose GPU-Accelerated Open-Source Suite for Multimodal Brain Data Visualization.

We present Visbrain, a Python open-source package that offers a comprehensive visualization suite for neuroimaging and electrophysiological brain data. Visbrain consists of two levels of abstraction: (1) objects which represent highly configurable neuro-oriented visual primitives (3D brain, sources connectivity, etc.) and (2) graphical user interfaces for higher level interactions. The object level offers flexible and modular tools to produce and automate the production of figures using an approach similar to that of Matplotlib with subplots. The second level visually connects these objects by controlling properties and interactions through graphical interfaces. The current release of Visbrain (version 0.4.2) contains 14 different objects and three responsive graphical user interfaces, built with PyQt: Signal, for the inspection of time-series and spectral properties, Brain for any type of visualization involving a 3D brain and Sleep for polysomnographic data visualization and sleep analysis. Each module has been developed in tight collaboration with end-users, i.e., primarily neuroscientists and domain experts, who bring their experience to make Visbrain as transparent as possible to the recording modalities (e.g., intracranial EEG, scalp-EEG, MEG, anatomical and functional MRI). Visbrain is developed on top of VisPy, a Python package providing high-performance 2D and 3D visualization by leveraging the computational power of the graphics card. Visbrain is available on Github and comes with a documentation, examples, and datasets (http://visbrain.org).

Money, power, glory: the linkages between EU conditionality and state capture in the Western Balkans

ABSTRACT The Western Balkan countries are stuck in a hybrid status quo on their way to democracy. Traditional arguments related to cost–benefit calculations, national identity or conflicting objectives fail to explain the observed decoupling between gradual improvements in formal compliance with membership criteria and stagnating, if not declining, democratic performance. We explain the limited impact of the EU’s political conditionality in the Western Balkans with rampant state capture and proceed to unpack how EU conditionality has effectively contributed to the consolidation of such detrimental governance patterns. First, EU pressure for simultaneous economic and political reforms opened opportunities for business actors to build powerful clientelist networks that reach into politics. Second, top-down conditionality has weakened political competition and mechanisms of internal accountability and deliberation. Finally, formal progress towards membership and high-level interactions with EU and member state officials legitimize corrupt elites. A congruence analysis of the Serbian case provides empirical evidence for the hypothesised linkages between EU conditionality and state capture.

Attention-based convolutional approach for misinformation identification from massive and noisy microblog posts

The fast development of social media fuels massive spreading of misinformation, which harm information security at an increasingly severe degree. It is urgent to achieve misinformation identification and early detection in social media. However, two main difficulties hinder the identification of misinformation. First, an event about a piece of suspicious news usually comprises massive microblog posts (hereinafter referred to as post), and it is hard to directly model the event with massive-volume posts. Second, information in social media is of high noise, i.e., most posts about an event have little contribution to misinformation identification. To resolve the difficulty of massive volume, we propose an Event2vec module to learn distributed representations of events in social media. To overcome the difficulty of high noise, we mine significant posts via content and temporal co-attention, which learn importance weights for content and temporal information of events. In this paper, we propose an Attention-based Convolutional Approach for Misinformation Identification (ACAMI) model. The Event2vec module and the co-attention contribute to learning a good representation of an event. Then the Convolutional Neural Network (CNN) can flexibly extract key features scattered among an input sequence and shape high-level interactions among significant features, which help effectively identify misinformation and achieve practical early detection. Experimental results on two typical datasets validate the effectiveness of the ACAMI model on misinformation identification and early detection tasks.

Assessment of Cognitive and Motor Skills in Parkinson's Disease by a Robotic Object Hitting Game.

Parkinson's disease (PD) patients experience various symptoms including extrapyramidal motor disturbances and cognitive impairments, which cause difficulties in daily life. However, PD patients have rarely been studied under realistic task situations that require high-level interaction of cognitive and motor skills. The aim of this study was to investigate the contribution of cognitive and motor factors to the performance of PD patients under high cognitive and kinematic loads. Twenty-six PD patients and 14 control subjects participated in the study. The PD patients performed a task involving hitting targets and avoiding distractors in levodopa On and Off states. A robotic manipulandum device recorded the numbers of target and distractor hits and hand kinematics, including movement area and speed. Performance on standard cognitive batteries and the Movement Disorder Society – Unified Parkinson's Disease Rating Scale motor scores were examined. The results indicated that the PD patients hit significantly fewer targets and more distractors than did the controls (p < 0.05). In PD patients, the average hand speed was slower and the area of hand movement was smaller than those of the control subjects (p < 0.001). Levodopa significantly increased the average hand speed and movement area (p < 0.01), but levodopa had an insignificant effect on the number of correct targets hit and erroneous distractor hits. The scores of cognitive batteries predicted the performance with regard to both targets hit and distractor avoidance. Our results were indicative of a dynamic interaction between cognitive and kinematic skills while the PD patients performed a virtual reality game. Single-dose levodopa enhanced kinematic capacity, and the global intelligence level predicted game performance.

Reviewing high-level control techniques on robot-assisted upper-limb rehabilitation

ABSTRACTThis paper presents a comprehensive review of high-level control techniques for upper-limb robotic training. It aims to compare and discuss the potentials of these different control algorithms, and specify future research direction. Included studies mainly come from selected papers in four review articles. To make selected studies complete and comprehensive, especially some recently-developed upper-limb robotic devices, a search was further conducted in IEEE Xplore, Google Scholar, Scopus and Web of Science using keywords (‘upper limb*’ or ‘upper body*’) and (‘rehabilitation*’ or ‘treatment*’) and (‘robot*’ or ‘device*’ or ‘exoskeleton*’). The search is limited to English-language articles published between January 2013 and December 2017. Valuable references in related publications were also screened. Comparative analysis shows that high-level interaction control strategies can be implemented in a range of methods, mainly including impedance/admittance based strategies, adaptive control techniques, and physiological signal control. Even though the potentials of existing interactive control strategies have been demonstrated, it is hard to identify the one leading to maximum encouragement from human users. However, it is reasonable to suggest that future studies should combine different control strategies to be application specific, and deliver appropriate robotic assistance based on physical disability levels of human users.

A Learning-Based Hierarchical Control Scheme for an Exoskeleton Robot in Human-Robot Cooperative Manipulation.

Exoskeleton robots can assist humans to perform activities of daily living with little effort. In this paper, a hierarchical control scheme is presented which enables an exoskeleton robot to achieve cooperative manipulation with humans. The control scheme consists of two layers. In low-level control of the upper limb exoskeleton robot, an admittance control scheme with an asymmetric barrier Lyapunov function-based adaptive neural network controller is proposed to enable the robot to be back drivable. In order to achieve high-level interaction, a strategy for learning human skills from demonstration is proposed by utilizing Gaussian mixture models, which consists of the learning and reproduction phase. During the learning phase, the robot observes and learns how a demonstrator performs a specific impedance-based task successfully, and in the reproduction phase, the robot can provide the subjects with just enough assistance by extracting human skills from demonstrations to prevent the motion of the robot end-effector deviating far from desired ones, due to variation in the interaction force caused by environmental disturbances. Experimental results of two different tasks show that the proposed control scheme can provide human subjects with assistance as needed during cooperative manipulation.

Persistent Stochastic Non-Interference

In this paper we present an information flow security property for stochastic, cooperating, processes expressed as terms of the Performance Evaluation Process Algebra (PEPA). We introduce the notion of Persistent Stochastic Non-Interference (PSNI) based on the idea that every state reachable by a process satisfies a basic Stochastic Non-Interference (SNI) property. The structural operational semantics of PEPA allows us to give two characterizations of PSNI: the first involves a single bisimulation-like equivalence check, while the second is formulated in terms of unwinding conditions. The observation equivalence at the base of our definition relies on the notion of lumpability and ensures that, for a secure process P, the steady state probability of observing the system being in a specific state P' is independent from its possible high level interactions.

Fitting the message to the location: engaging adults with antimicrobial resistance in a World War 2 air raid shelter.

There are many different initiatives, global and local, designed to raise awareness of antimicrobial resistance (AMR) and change audience behaviour. However, it is not possible to assess the impact of specific, small-scale events on national and international outcomes-although one might acknowledge some contribution to the individual and collective knowledge and experience-focused 'science capital' As with any research, in preparation for a public engagement event, it is important to identify aims, and appropriate methods whose results might help satisfy those aims. Therefore, the aim of this paper was to develop, deliver and evaluate an event designed to engage an adult audience with AMR. The venue was a World War 2 air raid shelter, enabling comparison of the pre- and postantibiotic eras via three different activity stations, focusing on nursing, the search for new antibiotics and investigations into novel antimicrobials. The use of observers released the presenters from evaluation duties, enabling them to focus on their specific activities. Qualitative measures of audience engagement were combined with quantitative data. The evaluation revealed that adult audiences can easily be absorbed into an activity-particularly if hands-on-after a brief introduction. This research demonstrates that hands-on practical engagement with AMR can enable high-level interaction and learning in an informal and enjoyable environment.

Improving the prediction of learning outcomes in educational platforms including higher level interaction indicators

AbstractOne of the most investigated questions in education is to know which factors or variables affect learning. The prediction of learning outcomes can be used to act on students in order to improve their learning process. Several studies have addressed the prediction of learning outcomes in intelligent tutoring systems environments with intensive use of exercises, but few of them addressed this prediction in other web‐based environments with intensive use not only of exercises but also, for example, of videos. In addition, most works on prediction of learning outcomes are based on low level indicators such as number of accesses or time spent in resources. In this paper, we approach the prediction of learning gains in an educational experience using a local instance of Khan Academy platform with an intensive use of exercises and taking into account not only low level indicators but also higher level indicators such as students' behaviours. Our proposed regression model is able to predict 68% of the learning gains variability with the use of six variables related to the learning process. We discuss these results providing explanation of the influence of each variable in the model and comparing these results with other prediction models from other works.

DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network

BackgroundMedical practitioners use survival models to explore and understand the relationships between patients’ covariates (e.g. clinical and genetic features) and the effectiveness of various treatment options. Standard survival models like the linear Cox proportional hazards model require extensive feature engineering or prior medical knowledge to model treatment interaction at an individual level. While nonlinear survival methods, such as neural networks and survival forests, can inherently model these high-level interaction terms, they have yet to be shown as effective treatment recommender systems.MethodsWe introduce DeepSurv, a Cox proportional hazards deep neural network and state-of-the-art survival method for modeling interactions between a patient’s covariates and treatment effectiveness in order to provide personalized treatment recommendations.ResultsWe perform a number of experiments training DeepSurv on simulated and real survival data. We demonstrate that DeepSurv performs as well as or better than other state-of-the-art survival models and validate that DeepSurv successfully models increasingly complex relationships between a patient’s covariates and their risk of failure. We then show how DeepSurv models the relationship between a patient’s features and effectiveness of different treatment options to show how DeepSurv can be used to provide individual treatment recommendations. Finally, we train DeepSurv on real clinical studies to demonstrate how it’s personalized treatment recommendations would increase the survival time of a set of patients.ConclusionsThe predictive and modeling capabilities of DeepSurv will enable medical researchers to use deep neural networks as a tool in their exploration, understanding, and prediction of the effects of a patient’s characteristics on their risk of failure.

AN ADAPTIVE PREDICTION OF AIRCRAFT MOTION WITH THE ELEMENTS OF VIRTUAL REALITY

The problem of adaptive method for predicting the aircraft state, which, by the dynamics control of the object, allows to reduce the load on the distributed network of virtual reality (with data consistency), while improving the accuracy of prediction the objects state, has been considered. The high-level protocol interaction approach let to organize interprocess communication network at a higher logical level, with less network loading, by supporting variable-length messaging and built-in mechanism the data replication control based on the principle “the election of consistency”. Method develops to reach better the prediction of the object rotation and trajectory motion correction of the aircraft.

Strategic Evacuation Network Design (SEND) under cost and time considerations

In this study, we pose and analyze an evacuation network design problem to provide a planning tool to help with high-level design decisions involved in strategic preparedness for large scale evacuations. In doing so, while incorporating evacuation time considerations, we also take a cost perspective in designing an effective evacuation network. Both the network design and the associated cost considerations in evacuation planning are commonly ignored in the literature due to a focus on evacuation time and the associated flow routing objective.We propose a mathematical model for Strategic Evacuation Network Design (SEND) that prescribes shelter regions and capacities, intermediate locations that support/supply for en route evacuees as well as road segments and their capacities under evacuation time constraints. To solve our model, we devise an efficient Benders Decomposition based approach enhanced with surrogate constraints, strengthened Benders cuts, heuristics, and the use of multi-cuts.We apply our methodology to solve test instances developed based on real data from Central Texas. We demonstrate by our analysis that the resulting approach does not only provide us with a means to design evacuation networks but also serves as a tool to study the trade-offs involved in design and operational performance measures as it captures the essence of high-level interactions between them.

Disentangling higher trophic level interactions in the cabbage aphid food web using high-throughput DNA sequencing

The lack of understanding of complex food-web interactions has been a major gap in the history of biological control. In particular, a better understanding of the functioning of pest food-webs and how they vary between native and invaded geographical ranges is of prime interest for biological control research and associated integrated pest management. Technical limitations associated with the deciphering of complex food-webs can now be largely overcome by the use of high throughput DNA sequencing techniques such as Illumina MiSeq. We tested the efficiency of this next generation sequencing technology in a metabarcoding approach, to study aphid food-webs using the cabbage aphid as model. We compared the variations in structure and composition of aphid food-webs in the species' native range (United Kingdom, UK) and in an invaded range (New Zealand, NZ). We showed that Illumina MiSeq is a well suited technology to study complex aphid food-webs from aphid mummies. We found an unexpectedly high top down pressure in the NZ cabbage aphid food-web, which coupled to a large ratio of consumer species / prey species and a lack of potential inter-specific competition between primary parasitoids, could cause the NZ food-web to be more vulnerable than the UK one. This study also reports for the first time the occurrence of a new hyperparasitoid species in NZ, as well as new associations between hyperparasitoids parasitoids and the cabbage aphid in this country. We conclude that the complexity of aphid food-webs in agricultural systems could often be underestimated, particularly at higher trophic levels; and that the use of high throughput DNA sequencing tools, could largely help to overcome this impediment.

8 Cognitive and psychiatric comorbidities in patients with vestibular disorder

Objective It is becoming increasingly clear from a variety of clinical and laboratory approaches that the human vestibular system not only regulates autonomic activity but also influences higher level processes. To help understand the nature of these higher-level interactions, we investigated the prevalence of cognitive, psychiatric and fatigue disturbances in a sample of 100 patients with primary vestibular disorder. The nature of this relationship and the variables influencing it were also explored. Method A comprehensive neuropsychological assessment battery, including measures of depression, anxiety, depersonalization, fatigue, sleep, memory, attention and executive function was completed by patients at their initial neuro-otology appointment. Neurological questionnaires and balance function tests were also administered. Results Unexpectedly, nearly all patients showed evidence of clinical impairment on most tests. Symptoms of fatigue, anxiety and working memory impairment were especially prevalent in the sample (over 50% fell outside the normative cut-off). Age, visual dominance, and previous access of psychological support were all significant predictors of the presence of these comorbidities. Conclusion These finding suggest that the vestibular system exerts a much wider influence on cognition and affect than previously indicated, and underlines the need for broader clinical diagnosis, assessment and management in patients who present with balance complaints.

PHIRST Trial – pharmacist consults: prioritization of HIV-patients with a referral screening tool

ABSTRACTThe role of pharmacists in HIV outpatient clinics has greatly increased in the past decades. Given the limited resources of the health system, the prioritization of pharmacist consults is now a main concern. This study aimed to create a scoring system allowing for standardized prioritization of pharmacist consults for patients living with HIV. Data was retrospectively collected from 200 HIV patients attending the Chronic Viral Illness Service at the McGill University Health Center. An expert panel consisting of four pharmacists working in the field of HIV prioritized each patient individually, after which a consensus was established and was considered as the gold standard. In order to create a scoring system, two different methods (Delphi, statistical) were used to assign a weight to each characteristic considered to be important in patient prioritization. A third method (equal weight to each characteristic) was also evaluated. The total score per patient for each method was then compared to the expert consensus in order to establish the score cut-offs to indicate the appropriate categories of delay in which to see the patient. All three systems failed to accurately prioritize patients into urgency categories (“less than 48 h”, “less than 1 month”, “less than 3 months”, “no consult required”) according to expert pharmacist consensus. The presence of high level interactions between patient characteristics, the limited number of patients and the low prevalence of some characteristics were hypothesized as the main causes for the results. Creating a prioritization tool for pharmacy consults in HIV outpatient clinics is a complex task and developing a decision tree algorithm may be a more appropriate approach in the future to take into account the importance of combinations of patient characteristic.

Adaptable scripting to foster regulation processes and skills in computer-supported collaborative learning

Collaboration scripts have repeatedly been implemented in Computer-Supported Collaborative Learning (CSCL) to facilitate collaboration processes and individual learning. However, finding the right degree of structure is a subtle design task: scripts that are too rigid may impair self-regulation and hinder learning; scripts that are too flexible may fail to evoke high-level interactions. This study investigated whether making collaboration scripts adaptable would be a way to raise their effectiveness. Three experimental phases were realized: In a first phase (exposure phase), all students solved three problem cases by aid of a collaboration script in an asynchronous, text-based CSCL environment. In a second phase (treatment phase), another three cases were presented that were to be solved by aid of a different theory that was presented to the learners through a summary on a sheet of paper. During this phase, a three-groups between-subject design was realized: (a) an unscripted condition, in which students received no specific guidance how to structure their collaboration, (b) a non-adaptable script condition, in which students’ collaboration was guided by the collaboration script they were trained in before, and (c) an adaptable script condition, in which students were allowed to modify parts of the trained script based on their self-perceived needs. In a third phase (subsequent transfer phase), students received a new case that they were to solve without guidance. N = 87 university students participated. Results showed that during the treatment phase, planning processes were most often performed in the unscripted condition. Yet, the adaptable script substantially increased students’ engagement in metacognitive activities of planning compared to learning with a non-adaptable script, and increased monitoring and reflection activities when compared to learning without script. Mediation analyses showed that the adaptable script facilitated learners’ use of self-regulation skills in the subsequent, unscripted transfer phase through the promotion of co-regulation processes of reflection in the treatment phase. The results reveal that adaptable scripting is a promising means of implementing flexible scripting and promoting self-regulation in CSCL.