Search

This article focused on two main topics currently on among many agendas: smart cities and Artificial Intelligence (AI). The growing interest in the former is due to these cities’ multidimensionality and adaptability in terms of residents’ needs and the requirements of each municipality's reality. AI, in turn, currently plays a transformative, disruptive role in various areas by performing “smart” tasks, thereby facilitating the automation of processes and differentiation initiatives. Smart city strategic planning, especially in the long term, will most likely have to deal with adaptations to AI. Thus, research is needed that can contribute to a more holistic view of these topics and support decision-making processes in these areas. Based on the epistemological principles of the multiple-criteria decision analysis approach (MCDA), this article developed and tested a dynamic analysis system that allows smart city initiatives to address the challenges of adapting to AI. The proposed system highlights the cause-and-effect relationships in this context. The article included an application of the decision-making trial and evaluation laboratory (DEMATEL) technique. The procedural steps followed to implement this methodology were enhanced by close collaboration with an experienced decision maker, who has coordinated various projects in this research context. The proposed system's contributions and limitations were also analyzed in this article.

Dynamic adaptation is an error-driven process of adjusting planned motor actions to changes in task dynamics (1). Adapted motor plans are consolidated into motor memories that contribute to better performance on re-exposure to the same dynamic condition. In parallel, dynamic perturbations can be compensated for by alternate motor control processes, such as co-contraction, that contribute to error reduction (2). Whether these control strategies share the same neural resources for memory formation is unclear. To address this gap in knowledge, we used a novel fMRI-compatible wrist robot, the MR-SoftWrist (3), to identify neural processes specific to dynamic adaptation and subsequent memory formation. Using the MR-SoftWrist, we acquired fMRI during a motor performance and a dynamic perturbation task to localize brain networks of interest. Resting state fMRI scans were acquired immediately before and after task performance to quantify changes in resting state functional connectivity (rsFC) within these networks. Twenty-four hours later, we assessed behavioral retention of training. A variance decomposition analysis was used to isolate behavior associated with adaptation versus alternate error reduction strategies. Immediately after the dynamic perturbation task, rsFC significantly increased within the corticothalamic-cerebellar network of the trained wrist and decreased interhemispherically within the cortical sensorimotor network. These changes were associated to behavioral measures of initial acquisition and retention, indicative of memory formation. Variance decomposition analysis revealed that increases within the cortico-thalamic-cerebellar network were associated with adaptation, while interhemispheric decreases in rsFC within the sensorimotor network were associated with alternate error reduction processes.Significance StatementMotor memory formation processes have been studied using resting state functional connectivity (rsFC) before and after exposure to adaptation tasks. However, due to technical limitations, previous studies have not investigated rsFC within brain regions localized during task execution, nor quantified rsFC within minutes after task performance. The present study used an fMRI-compatible wrist robot to localize relevant brain regions during dynamic adaptation within the cortico-thalamic-cerebellar network and bilateral sensorimotor network. Immediately following dynamic adaptation, we measured significant changes in rsFC within these networks, that were associated with adaptation behavior and with gains in behavior assessed 24 hours later, indicative of memory formation. Variance decomposition analysis identified distinct rsFC networks associated with adaptation specific processes, and with alternate motor control strategies.

Based on advanced information and communication infrastructures and enabled with the cutting-edge information processing of cognitive computing, existing smart manufacturing systems have foreseen a prevailing tendency that approaches a higher automation level, i.e., Self-X (e.g., Self-configuration/optimization/adaptation). However, the readiness of ‘Self-X’ levels is still far to reach, encountering the practical challenges of semantics-based networking and human-machine untrust in the manufacturing scenario. To mind these gaps, the authors envision an industrial knowledge graph (IKG) and graph embedding (GE) enabled pathway, to flourish today’s smart manufacturing paradigms towards cognitive mass personalization. To pave it, three promising IKG and GE enabling techniques in the ‘Self-X’ cognitive manufacturing network are described. Potential opportunities and challenges are also pointed out to invite more opinions to refine and innovate the exploitation of IKG and GE for the future of smart manufacturing.

Cel pracyCelem badania jest przedstawienie polskiej wersji Metody Oceny Wypalenia Zawodowego BAT-PL W. Schaufeliego i wsp. oraz oceną jej trafności i rzetelności. Metoda mierzy podstawowe objawy wypalenia (BAT-C): wyczerpanie, zdystansowanie psychiczne, pogorszenie funkcjonowania poznawczego i emocjonalnego, oraz towarzyszące im objawy wtórne (BAT-S): skargi psychosomatyczne i objawy dystresu psychologicznego.MetodaW badaniu uczestniczyło 255 osób personelu pielęgniarskiego. W celu oceny trafności wykorzystano jednopunktową skalę satysfakcji z pracy, Utrechtską Skalę Zaangażowania w Pracę UWES-3 Schaufeliego i wsp. oraz Skalę Dobrostanu Emocjonalnego w Pracy JAWS van Katwyk i wsp.WynikiWyniki konfirmacyjnej analizy czynnikowej potwierdziły strukturę czynnikową metody. Dotyczy to zarówno BAT-C jako czterech objawów podstawowych i czynnika ogólnego (metacecha) reprezentującego wypalenie, jak i BAT-S jako zespołu dwóch objawów wtórnych oraz metacechy. Obie skale były silnie skorelowane ze sobą i różniły się od innych miar dobrostanu zawodowego (satysfakcji z pracy, zaangażowania w pracę i emocji negatywnych). Wartości współczynników α-Cronbacha oraz rzetelności łącznej pokazały, że BAT jest rzetelną metodę pomiaru.WnioskiPolska adaptacja Metody Oceny Wypalenia Zawodowego BAT-PL W. Schaufeliego i wsp. cechuje się dobrymi właściwościami psychometrycznymi i może być brana pod uwagę w badaniach naukowych nad wypaleniem zawodowym oraz dalej weryfikowana w praktyce klinicznej.

Adaptive behavior requires learning about the structure of one’s environment to derive optimal action policies, and previous studies have documented transfer of such structural knowledge to bias choices in new environments. Here, we asked whether people could also acquire and transfer more abstract knowledge across different task environments, specifically expectations about cognitive control demands. Over three experiments, participants performed a probabilistic card-sorting task in environments of either a low or high volatility of task rule changes (requiring low or high cognitive flexibility respectively) before transitioning to a medium-volatility environment. Using reinforcement learning modeling, we consistently found that previous exposure to high task rule volatilities led to faster adaptation to rule changes in the subsequent transfer phase. These transfers of expectations about cognitive flexibility demands were both task- (Experiment 2) and stimulus- (Experiment 3) independent, thus demonstrating the formation and generalization of environmental structure knowledge to guide cognitive control.Statement of RelevanceWe investigated whether structural knowledge of one task environment can be transferred to guide cognitive control strategies in new environments. Past research has found that while learning generally improves subsequent performance, it does so only for the learned task (“near transfer”) and has little or no generalizability to novel task rules and stimuli (“far transfer”). However, recent studies suggest that learning more abstract, structural task features (e.g., cognitive maps) allows for that knowledge to be applied to new environments. Here, we took a critical additional step and showed that people can acquire and transfer expectations about cognitive control demands (specifically cognitive flexibility) across different task environments. To our knowledge, this is the first demonstration of people’s ability to extract and re-use cognitive control learning parameters that transcend specific stimuli and tasks. This transfer of learned cognitive flexibility is particularly noteworthy because such flexibility is impaired in several common psychiatric conditions.

Integrating Hobfoll's (1989) conservation of resources theory with Berry's (1997) acculturation taxonomies, we develop a typological theory to explain the acculturation stress and adaptation proces...

Unsupervised domain adaptation for person re-identification (Re-ID suffers severe domain discrepancies between source and target domains. To reduce the domain shift caused by the changes of context, camera style, or viewpoint, existing methods in this field fine-tune and adapt the Re-ID model with augmented samples, either through translating source samples to the target style or by assigning pseudo labels to the target. The former methods may lose identity details but keep redundant source background during translation, while the latter methods may assign noisy labels when the model meets the unseen background and person pose. To address the challenges, we mitigate the domain shift in the former translation direction by decoupling environment and identity-related features in a cyclic manner. We propose a novel individual-preserving and environmental-switching cyclic generation network (IPES-GAN . Our network has the following distinct features: 1 Decoupled features instead of fused features: the images are encoded into an individual part and an environmental part, which are proved beneficial to generation and adaptation; 2 Cyclic generation instead of one-step adaptive generation. The source and target environment features are swapped to generate cross-domain images with preserved identity-related features conditioned with source (target environment features, and then swapped again to generate back the input image, so that cyclic generation runs in a self-supervised way. Experiments carried out on two major benchmarks: Market-1501 and DukeMTMC-reID reveal state-of-the-art performance.

Industry 4.0 is a revolutionizing concept which has a more significant implication to numerous industries of various sectors worldwide. Industry 4.0 acts as a technical catalyst in accelerated growth through systematic adaptation and new-generation cutting-edge engineering tools. IoT, Blockchain, Artificial Intelligence, Augmented Reality, 3D printing, Big-data analytics, cloud-computing are some of the primary technologies which work in join hand to make the vision of Industry 4.0 successful. In addition, it promotes autonomous industry, which in turn considerably reduces human intervention, thus resulting in efficient working and safeguard companies' interests. Due to the faster adaption of Industry 4.0, authors have attempted to define each and every concept related to it and to explore a futuristic roadmap, special focusing on its implication in the contemporary market. In the 21st century Industry, 4.0 is the need of the hour, so it has become essential to understand the state-of-the-art technologies involved.

It is widely acknowledged that the success of deep learning is built on large-scale training data and tremendous computing power. However, the data and computing power are not always available for many real-world applications. In this paper, we address the machine learning problem where it lacks training data and limits computing power. Specifically, we investigate domain adaptation which is able to transfer knowledge from one labeled source domain to an unlabeled target domain, so that we do not need much training data from the target domain. At the same time, we consider the situation that the running environment is confined, e.g., in edge computing the end device has very limited running resources. Technically, we present the Faster Domain Adaptation (FDA) protocol and further report two paradigms of FDA: early stopping and amid skipping. The former accelerates domain adaptation by multiple early exit points. The latter speeds up the adaptation by wisely skip several amid neural network blocks. Extensive experiments on standard benchmarks verify that our method is able to achieve the comparable and even better accuracy but employ much less computing resources. To the best of our knowledge, there are very few works which investigated accelerating knowledge adaptation in the community.

Video frame interpolation is a challenging problem that involves various scenarios depending on the variety of foreground and background motions, frame rate, and occlusion. Therefore, generalizing across different scenes is difficult for a single network with fixed parameters. Ideally, one could have a different network for each scenario, but this will be computationally infeasible for practical applications. In this work, we propose MetaVFI, an adaptive video frame interpolation algorithm that uses additional information readily available at test time but has not been exploited in previous works. We initially show the benefits of test-time adaptation through simple fine-tuning of a network and then greatly improve its efficiency by incorporating meta-learning. Thus, we obtain significant performance gains with only a single gradient update without introducing any additional parameters. Moreover, the proposed MetaVFI algorithm is model-agnostic which can be easily combined with any video frame interpolation network. We show that our adaptive framework greatly improves the performance of baseline video frame interpolation networks on multiple benchmark datasets.

Military technologies are typically the product of long-term development efforts. These technologies are often adapted late in the development process for use by other users, including other members of the defense enterprise, foreign militaries, or the commercial sector. The need to adapt a technology can arise from cost over-runs or changing operational requirements. This article describes a value-based methodology for developing a transition plan early in the development process, when the system is still in the conceptual phase. Performing this analysis early in the design phase allows for design choices that will support adapting the system for other users. Moreover, it provides a risk mitigation strategy against cost over-run and changes in operational requirements. The process begins by capturing the functional architecture of the technology and analyzing them against a set of tasks associated with different military positions. From there, the adapted technology is evaluated for its projected adequacy based on the value added and the effort required for adaptation. The end-state of this analysis is a value-based model that identifies a portfolio of possible alternative markets and/or uses, which can be comparatively analyzed. A case study is presented for adapting a military augmented reality system for a different market.

The review outlines the milestones in studying the processes of human adaptation to hypoxia, and hypoxic training applications in medicine and sports. Contribution of the Russian science to these investigations is disclosed and literary data on the mechanisms of hypoxic adaptation, models and effectiveness of hypoxic training are summarized. The paper is concluded by discussion of hypoxic training potential in high achievements sports.

The article considers the adaptive behavior of Russians during 2015–2017 (a period that is often referred to as the “new reality”) by analyzing how Russians perceive the prospects of economic devel...

Many applications utilize sensors on mobile devices and apply deep learning for diverse applications. However, they have rarely enjoyed mainstream adoption due to many different individual conditions users encounter. Individual conditions are characterized by users unique behaviors and different devices they carry, which collectively make sensor inputs different. It is impractical to train countless individual conditions beforehand and we thus argue meta-learning is a great approach in solving this problem. We present MetaSense that leverages seen conditions in training data to adapt to an unseen condition (i.e., the target user). Specifically, we design a meta-learning framework that learns how to adapt to the target via iterative training sessions of adaptation. MetaSense requires very few training examples from the target (e.g., one or two) and thus requires minimal user effort. In addition, we propose a similar condition detector (SCD) that identifies when the unseen condition has similar characteristics to seen conditions and leverages this hint to further improve the accuracy. Our evaluation with 10 different datasets shows that MetaSense improves the accuracy of state-of-the-art transfer learning and meta learning methods by 15% and 11%, respectively. Furthermore, our SCD achieves additional accuracy improvement (e.g., 15% for human activity recognition).

Mobile Deep Learning (MDL) has emerged as a privacy-preserving learning paradigm for mobile devices. This paradigm offers unique features such as privacy preservation, continual learning and low-latency inference to the building of personal mobile sensing applications. However, squeezing Deep Learning to mobile devices is extremely challenging due to resource constraint. Traditional Deep Neural Networks (DNNs) are usually over-parametered, hence incurring huge resource overhead for on-device learning. In this paper, we present a novel on-device deep learning framework named MDLdroidLite that transforms traditional DNNs into resource-efficient model structures for on-device learning. To minimize resource overhead, we propose a novel Release-and-Inhibit Control (RIC) approach based on Model Predictive Control theory to efficiently grow DNNs from tiny to backbone. We also design a gate-based fast adaptation mechanism for channel-level knowledge transformation to quickly adapt new-born neurons with existing neurons, enabling safe parameter adaptation and fast convergence for on-device training. Our evaluations show that MDLdroidLite boosts on-device training on various PMS datasets with 28x to 50x less model parameters, 4x to 10x less floating number operations than the state-of-the-art model structures while keeping the same accuracy level.

Unsupervised domain adaptation (UDA) aims at learning a classifier for an unlabeled target domain by transferring knowledge from a labeled source domain with a related but different distribution. Most existing approaches learn domain-invariant features by adapting the entire information of the images. However, forcing adaptation of domain-specific variations undermines the effectiveness of the learned features. To address this problem, we propose a novel, yet elegant module, called the deep ladder-suppression network (DLSN), which is designed to better learn the cross-domain shared content by suppressing domain-specific variations. Our proposed DLSN is an autoencoder with lateral connections from the encoder to the decoder. By this design, the domain-specific details, which are only necessary for reconstructing the unlabeled target data, are directly fed to the decoder to complete the reconstruction task, relieving the pressure of learning domain-specific variations at the later layers of the shared encoder. As a result, DLSN allows the shared encoder to focus on learning cross-domain shared content and ignores the domain-specific variations. Notably, the proposed DLSN can be used as a standard module to be integrated with various existing UDA frameworks to further boost performance. Without whistles and bells, extensive experimental results on four gold-standard domain adaptation datasets, for example: 1) Digits; 2) Office31; 3) Office-Home; and 4) VisDA-C, demonstrate that the proposed DLSN can consistently and significantly improve the performance of various popular UDA frameworks.

Introducción: Este artículo describe la planificación efectuada para adecuar la capacidad diagnóstica de grandes volúmenes de RT-PCR de SARS-CoV-2. Métodos: El análisis y predicción del flujo de trabajo incluyó el número de RT-PCR desde el inicio de la pandemia, con 31971 registros. La planificación de la capacidad y de las opciones diagnósticas se planteó en base a los posibles escenarios derivados de las predicciones efectuadas. Resultados: En base a las predicciones obtenidas se optó por una solución automatizada basada en el empleo de robots OT-2 (Opentrons) para configurar un flujo de trabajo reproducible, que logró una capacidad de procesamiento de 5640 muestras/día, con un tiempo de respuesta de 4 horas. Conclusiones: El análisis y predicción del flujo de trabajo, unido al empleo de plataformas basadas en OT-2, proporciona una infraestructura robusta que permite atender con éxito las demandas de prueba que exige esta pandemia.

Teachers’ instructional practices are important for student learning. However, there are few tools for evaluating instructional practices in social studies. To this end, we present a video-based instrument with teacher-student interactions for assessing teaching practices in a high school economics course. We investigate, using an iterative design-based methodology, the adaptation and validation of a test originally developed in Germany for use in the United States. We found that experienced economics teachers and those with more content knowledge demonstrated higher-level instructional skills, and were more flexibile with their thinking. Inexperienced economics teachers and those with low content knowledge not only struggled with the test items, but also with the test format, the content, inter-cultural understanding, and with feeling overwhelmed. The test provides an opportunity for economics teachers to obtain feedback, practice, reflection, and improvement of their instruction in a way that has not previously been available. It also illustrates the possibilities and drawbacks of video-based testing in teacher training.

Abstract. This research provides a systematic overview of psychological areas using assimilation and accommodation to explain development and adaptation processes from 1998 to 2018. We primarily aimed to identify the main psychological research areas connected to assimilation and accommodation. We used assimilation and accommodation as keywords to extract data from SpringerLink, PsycINFO, and PsycARTICLES. Of 500 articles, 473 were included in the analysis. Ten categories were identified to allow systematization along with different research areas and development trajectories. The meanings of these terms were analyzed in terms of scientific impact, their connection to Piaget and Baldwin, application, and research methods. Our analysis has distilled the most driving and scientifically relevant approaches to assimilation and accommodation within psychological research, with the work of Baldwin and Piaget influencing practically all views. Thus, we have identified a common understanding of assimilation and accommodation, although the direction of the adaptation process should be made explicit in the future. Based on our analyses, we were able to identify white spots on the research map that should be focused on in future work: the need to better understand the interdependence and synchronicity of both processes, the connection to affects and emotions, and the potential co-research with artificial intelligence.

We propose a learning-based model for multifunctional elbow exoskeleton control, i.e., assist- and resist-as-needed (AAN and RAN). The model consists of online iterative learning and impedance adaptation mechanisms for predictive and variable compliant joint control. The model was implemented on a lightweight (0.425 kg) and portable elbow exoskeleton (i.e., POW-EXO) worn by three subjects, respectively. The implementation relies only on internal pose (e.g., joint position) feedback, rather than physical compliant mechanisms (e.g., springs) and external sensors (e.g., electromyography (EMG) or force) typically required by conventional exoskeletons and controllers. The proposed model provides a novel technique to achieve multifunctional exoskeleton control with minimal mechatronics and sensing. Interestingly, its RAN control and POW-EXO as a quantification means, may reveal interactive (mechanical) impedance variance and invariance in human motor control.

Seizure prediction from intracranial electroencephalogram (iEEG) has great potentials to improve the life quality of epileptic patients, but faces big challenges. One major difficulty lies in that brain signal changes occasionally during long-term monitoring, due to electrode movements or the nonstationary brain dynamics. This leads to a serious situation that a predictor learned from historical data usually only works well in a short time period as long as the data do not change much. While in a long time span, the performance of the learned features decreases or even become totally invalid. To deal with the problem, we propose a domain adaptation convolutional neural network to learn robust preictal features that is invariant across different time periods. Specifically, the preictal feature is learned and enhanced by multi-scale temporal convolutions in the neural network. Based on this, a domain adaptation method is adopted to constrain that the learned features should be invariant across different time periods. Experimental results demonstrate that our approach can effectively improve seizure prediction performance against signal changes.

Deep neural networks (DNNs) have been widely used in many Cyber-Physical Systems (CPS). However, it is still challenging work to deploy DNNs in real-time systems. In particular, the execution time of DNN inference must be predictable, s.t. it could be known whether the run-time inference can complete within a required timing constraint. Moreover, the timing constraints may change dynamically with the run-time environment in many embedded applications, such as autonomous cars. A possible way to meet such dynamic real-time requirements is to execute different sub-networks of a DNN at run-time. However, improper construction of sub-networks may not only introduce unpredictable inference time, s.t. the real-timing constraints could be violated unexpectedly, but also has poor compatibility with the well-optimized machine learning framework (e.g., TensorFlow). In this paper, we study the predictability when executing different sub-networks of a DNN. In particular, we present a feature-wise run-time adaptation framework for DNN inference, which is implemented and validated on NVIDIA Jetson TX2 and Nano with TensorFlow. The experimental results show that our method can achieve predictable inference time in comparison with the state-of-the-art methods.

Heterogeneous bounded confidence opinion dynamics with an adaptation policy of the agents' confidence thresholds is considered. The strategy implemented by each agent consists of increasing his confidence thresholds when he has no active neighbor, thus strengthening his heterophilous propensity. Moreover, his adaptation algorithm is stopped if there is a minimal number of agents within his similarity interval. The heterogeneity of the model could lead to interesting scenarios, e.g. steady-state oscillatory behaviors and practical clustering. Sufficient conditions for reaching in finite time a practical consensus are proposed. An upper bound on the maximum number of steady-state practical clusters which depends on the parameters of the confidence thresholds adaptation algorithm is obtained. The steady-state behaviors in the presence of one-sided confidence thresholds and stubbornness are discussed. Numerical simulations verify the theoretical results.

The internet-of-things (IoT) refers to the growing number of embedded interconnected devices within everyday ubiquitous objects and environments, especially their networks, edge controllers, data gathering and management, sharing, and contextual analysis capabilities. However, the IoT suffers from inherent limitations in terms of human-computer interaction. In this landscape, there is a need for interfaces that have the potential to translate the IoT more solidly into the foreground of everyday smart environments, where its users are multimodal, multifaceted, and where new forms of presentation, adaptation, and immersion are essential. This work highlights the synergetic opportunities for both IoT and XR to converge toward hybrid XR objects with strong real-world connectivity, and IoT objects with rich XR interfaces. The paper contributes i) an understanding of this multi-disciplinary domain XR-IoT (XRI); ii) a theoretical perspective on how to design XRI agents based on the literature; iii) a system design architectural framework for XRI smart environment development; and iv) an early discussion of this process. It is hoped that this research enables future researchers in both communities to better understand and deploy hybrid smart XRI environments.

A fuzzy adaptive switching event-triggered control strategy is obtained for a class of uncertain nonlinear switched systems with predefined accuracy. Since the switching may occur between two adjacent sampling instants and may lead to trigger a new event, it raises the mismatched switching and brings severe challenges to avoid the Zeno behavior of the subsystem controllers. The formidable problems are successfully addressed by a new switching event-triggering mechanism without restrictive assumptions. With the proposed fuzzy adaptive switching control approach, the adverse effects of mismatched switching between the subsystem and the controller are addressed, and no subsystem controllers exhibit the Zeno behavior. Furthermore, by using a projection operator-based adaptation strategy and constructing a class of continuous functions, the convergence accuracy of the switched system appears explicitly in design parameters. It allows designers to make decisions in advance based on the requirement of convergence accuracy. Finally, theoretical results are verified by two illustrative examples.

Real-world infrastructures modeled as complex networks have been popular topics in recent decades. As these infrastructures develop, becoming critical to support millions of people, their resilience in the face of disasters and attacks becomes an increasing concern to the engineers and stakeholders. Using a graph theory and complex network approach, we conducted an empirical study on the resilience of five metro, or subway systems selected globally. A multi-phase resilience framework was implemented from the resilience cycle concept, including preparedness, robustness, recovery, and adaptation. It also incorporated the historical flow data into the resilience analysis to account for user behavior impacts. The paper demonstrates a systematic comparison of five metro networks and explores trends in each resilience phase. Generally, diverse performances were observed among the five metro networks, with the Singapore metro showing the best resilience. It is robust to both random disruption and targeted attacks. Analysis suggests its advantage may come from both its moderate heterogeneous topology and the flow pattern, exhibiting the smallest average travel distance. Besides, the result highlights the significance of commuters' relocation behavior between neighboring stations in the recovery stage. For the adaptation from a long-term perspective, we developed a screening tool to investigate the impact of the network expansion on the existing part of network by mapping its relationship with the characteristics of the newly connected nodes. In general, the observation of the networks' different behaviors in flow-weighted analysis symbolizes that flow is necessary for consideration in the resilience design of real-world transportation systems.

A central capability of a long-lived reinforcement learning (RL) agent is to incrementally adapt its behavior as its environment changes, and to incrementally build upon previous experiences to facilitate future learning in real-world scenarios. In this paper, we propose LifeLong Incremental Reinforcement Learning (LLIRL), a new incremental algorithm for efficient lifelong adaptation to dynamic environments. We develop and maintain a library that contains an infinite mixture of parameterized environment models, which is equivalent to clustering environment parameters in a latent space. The prior distribution over the mixture is formulated as a Chinese restaurant process (CRP), which incrementally instantiates new environment models without any external information to signal environmental changes in advance. During lifelong learning, we employ the expectation maximization (EM) algorithm with online Bayesian inference to update the mixture in a fully incremental manner. In EM, the E-step involves estimating the posterior expectation of environment-to-cluster assignments, while the M-step updates the environment parameters for future learning. This method allows for all environment models to be adapted as necessary, with new models instantiated for environmental changes and old models retrieved when previously seen environments are encountered again. Experiments demonstrate that LLIRL outperforms relevant existing methods, and enables effective incremental adaptation to various dynamic environments for lifelong learning., Comment: Accepted by IEEE Transactions on Neural Networks and Learning Systems, 2021

Differences between military and civilian culture can impact a veteran's ability to effectively navigate the transition to the civilian setting after separating from military service. However, systems providing psychological services to veterans lack reliable and valid methods of identifying the extent to which the dissonance between military and civilian cultures is associated with postseparation adjustment (e.g., ability to integrate profound deployment experiences into civilian life) and psychological outcomes. Utilization of a theory-driven, acculturation framework to assess military and civilian cultural affiliation may address this gap. While several quantitative measures exist for evaluating acculturation in reference to different ethnocultural experiences and that evaluate military culture-related adjustment or reintegration, there are no existing validated measures rooted in cultural theory that explicitly assess military-related acculturation. The aim of the present study was to develop and examine the psychometric properties of theoretically grounded, military-related adaptations of validated acculturation measures. Data from 364 veterans were collected with an online survey including four military-related adaptations of acculturation measures. Results suggest that the proposed adaptation of the original ethnocultural measurement model was not a good fit to the data, χ² = 2396.99, p < .001, TLI = .73. An evaluation of localized areas of strain in the confirmatory model and results from an exploratory factor analysis suggest support for the continued development of a military-related model of acculturation after considering further measure restructuring and validation. Opportunities for future research and measure validation are discussed. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Object detection in traffic scenes has attracted considerable attention from both academia and industry recently. Modern detectors achieve excellent performance under a simple constrained environment while performing poorly under the actual complex and open traffic environment. Therefore, the capability of adapting to new and unseen domains is a key factor for the large-scale application and proliferation of detectors in autonomous driving. To this end, this paper proposes a novel category-induced coarse-to-fine domain adaptation approach (C2FDA) for cross-domain object detection, which consists of three pivotal components: (1) Attention-induced coarse-grained alignment module (ACGA), which strengthens the distribution alignment across disparate domains within the foreground features in category-agnostic way by the minimax optimization between the domain classifier and the backbone feature extractor; (2) Attention-induced feature selection module, which assists the model to emphasize the crucial foreground features and enables the ACGA to focus on the relevant and discriminative foreground features, without being affected by the distribution of inconsequential background features; (3) Category-induced fine-grained alignment module (CFGA), which reduces the domain shift in category-aware way by minimizing the distance of centroids with the same category from different domains and maximizing that of centroids with disparate categories. We evaluate the performance of our approach in various source/target domain pairs and comprehensive results demonstrate that C2FDA significantly outperforms the state-of-the-art on multiple domain adaptation scenarios, i.e., the synthetic-to-real adaptation, the weather adaptation, and the cross camera adaptation.

Due to the data distribution discrepancy, fault diagnosis models, trained with labeled data in one scene, likely fails in classifying by unlabeled data acquired from the other scenes. Transfer learning is capable to generalize successful application trained in one scene to the fault diagnosis in the other scenes. However, the existing transfer methods do not pay much attention to reduce adaptively marginal and conditional distribution biases, and also ignore the degree of contribution between both biases and among network layers, which limit classification performance and generalization in reality. To overcome these weaknesses, we established a new fault diagnosis model, called subdomain adaptation transfer learning network (SATLN). Firstly, two convolutional building blocks were stacked to extract transferable features from raw data. Then, the pseudo label learning was amended to construct target subdomain of each class. Furthermore, a sub-domain adaptation was combined with domain adaptation to reduce both marginal and conditional distribution biases simultaneously. Finally, a dynamic weight term was applied for adaptive adjustment of the contributions from both discrepancies and each network layers. The SATLN method was tested with six transfer tasks. The results demonstrate the effectiveness and superiority of the SATLN in the cross-domain fault diagnosis field.

An active, compact, wideband, receiving filtenna with power adaptation for space-limited wireless platforms is presented. By organically combining a filtering antenna and a Variable Gain Amplifier (VGA), the proposed co-designed active filtenna is achieved. Its power adaptation is empowered by controlling gain of the VGA unit. A demonstrated prototype with a compact overall size is fabricated, assembled and measured. Good performances are achieved. Furthermore, in any condition, the proposed active filtenna displays outstanding impedance matching (|S11

The aim of steganography detection is to identify whether the multimedia data contain hidden information. Although many detection algorithms have been presented to solve tasks with inconsistent distributions between the source and target domains, effectively exploiting transferable correlation information across domains remains challenging. As a solution, we present a novel multiperspective progressive structure adaptation (MPSA) scheme based on active progressive learning (APL) for JPEG steganography detection across domains. First, the source and target data originating from unprocessed steganalysis features are clustered together to explore the structures in different domains, where the intradomain and interdomain structures can be captured to provide adequate information for cross-domain steganography detection. Second, the structure vectors containing the global and local modalities are exploited to reduce nonlinear distribution discrepancy based on APL in the latent representation space. In this way, the signal-to-noise ratio (SNR) of a weak stego signal can be improved by selecting suitable objects and adjusting the learning sequence. Third, the structure adaptation across multiple domains is achieved by the constraints for iterative optimization to promote the discrimination and transferability of structure knowledge. In addition, a unified framework for single-source domain adaptation (SSDA) and multiple-source domain adaptation (MSDA) in mismatched steganalysis can enhance the model's capability to avoid a potential negative transfer. Extensive experiments on various benchmark cross-domain steganography detection tasks show the superiority of the proposed approach over the state-of-the-art methods.

A survey of the field of Interlanguage Pragmatics (ILP) shows that the Discourse Completion Test (DCT)2, also referred to as a ‘discourse completion task’ or a ‘production questionnaire’, has been the most frequently used instrument to evaluate second/foreign language learners’ ability to perform speech acts in a target language, despite the harsh criticism leveled against its low construct validity and its failure to represent the features of authentic discourse. Interestingly, focusing on the statement of objectives of a number of ILP studies using DCTs, one can notice that such studies rarely refer to the DCT as a language test. In addition, an overview of the DCT design process as described in several ILP studies shows that ever since its adaptation for the study of pragmatic ability (Blum-Kulka, 1982), there has been a tendency to use or adapt one of the existing DCT versions used in previous studies based on the argument of comparability of results. While a number of ILP researchers tried to improve the design of the DCT by the inclusion of rejoinders or by enhancing the prompt material (e.g. Billmyer and Varghese, 2000), few attempts have been made to reconsider the DCT development process. McNamara and Roever (2006: 253) urge for the need for “more research on testing of sociopragmatic knowledge and design of discourse completion tests for testing purposes.”The present paper starts with an overview of the literature about DCTs with special reference to the cognitive validity of the instrument and to previous studies dealing with DCT structure and content. Then, with reference to research in the fields of language testing and psychometrics, it shows that, whether used for research or instructional purposes, the DCT shares several qualities with language tests. As such, it is argued that the DCT should be treated as a language test and not as a questionnaire and should, thus, undergo a rigorous developmental process. Based on recent models of language test construction, the paper concludes with an overview of the stages of DCT development.

The study of language contact has been traditionally carried out from a structural perspective (synchronic or diachronic), from a sociolinguistic perspective and/or from a rather psychological perspective, centered on the linguistic and communicative competence of the multilingual individual. However, a great number of linguistic and sociolinguistic topics that appear in language contact situations may be productively tackled from a pragmatic viewpoint. This pragmatic perspective takes into account linguistic use in communication contexts and raises, at a different level, questions that deal with the structures and the evolution of the codes in contact. The main aim of this presentation is the analysis of some of the specific problems that arise in given language contact situations from a pragmatic perspective, considering the adaptation processes of the speakers, their particular interactive strategies and the social meaning generated. Understanding pragmatics in its original sense, i.e. as the study of the relationship between linguistic signs and speakers (users of certain resources), these phenomena should be understood as the result of speakers’ adaptation to changing sociocultural circumstances. This adaptation creates a new distribution of the verbal resources (or linguistic economy) of the community and, consequently, modifies its varieties as far as form and function are concerned.

The paper examines the pragmatic roles that locutionary acts play in understanding the communication between doctors and patients in Southwestern Nigeria. Working within John Austin’s locutionary acts, with restrictions to the lexical occurrences and lexical relationships observed in the discourse, it got data through tape recordings of doctor-patient conversations and interviews of both doctors and patients (and/or their relations).The findings revealed that two categories of locutions were engaged in hospital interactions, namely, locutions intended to be understood by non-professionals and locutions not intended to be understood by non-professionals. The paper observes that locutions in medical discourse in Southwestern Nigeria bring standard lexical choices and local linguistic initiatives of medical practitioners into a pragmatic union. It therefore concludes that the pragmatic engagement of these choices displays the tact the practitioners use in dealing with patients, and it recommends the need for the practitioners to master the locutions and their pragmatic adaptation for effective management of patients.

Most robotic systems designed for mass manufacturing are optimized for a specific type of product. They generally lack the ability to adapt to low-volume customized products. In this article, we present a system based on a modular design for manufacturing personalized medical stent graft implants. The concept is based on learning-by-demonstration by integrating real-time 3-D vision, multirobot collaboration, and personalization to guide the robots to learn and execute tasks continuously with adaptation to different implant geometry. The system is optimized to generate customized and collision-free paths for efficient object manipulation and task completion. We show that the system is generalizable to different stent graft designs and the proposed multirobot system can seamlessly work together with high efficiency without collisions. The results have also suggested its usability for other manipulation tasks, especially for flexible production of customized products where bimanual or multirobot cooperation is required.

OBJECTIVE: To validate and refine typologies of consultation performance from previous research to identify learning needs associated with each typology.METHODS: We performed a qualitative study in a General Practice Specialty Training programme, using a two-stage design. First, we selected four exemplars from 80 videotaped consultations of 7 first-year and 6 third-year trainees that reflected the four typologies. We subsequently held individual interviews with clinical supervisors (N = 20) who observed these consultations to identify recurrent trainee behaviours.RESULTS: The 'doctor-patient interaction' dimension from previous research was specified to encompass relationship-building, exploring, structuring, and shared decision-making competencies. Medical expertise was a moderating factor. The attitude and consultation behaviours included in the typologies were validated and we formulated directions for learning based on learning needs identified per typology.CONCLUSION: Supervisors have a shared frame of reference for the behaviours reflecting proficient consultation performance. Serving as a developmental road map, all learning needs emphasised contextual adaptation, calling for an improved balance between patient-centred relationship building and application of medical expertise.PRACTICE IMPLICATIONS: By providing rich and tailored feedback on consultation performance, the refined typologies - albeit subject to additional refinement in future research - may promote the monitoring of individual competence development over time.

Although speech driven facial animation has been studied extensively in the literature, works focusing on the affective content of the speech are limited. This is mostly due to the scarcity of affective audio-visual data. In this article, we improve the affective facial animation using domain adaptation by partially reducing the data scarcity. We first define a domain adaptation to map affective and neutral speech representations to a common latent space in which cross-domain bias is smaller. Then the domain adaptation is used to augment affective representations for each emotion category, including angry, disgust, fear, happy, sad, surprise, and neutral, so that we can better train emotion-dependent deep audio-to-visual (A2V) mapping models. Based on the emotion-dependent deep A2V models, the proposed affective facial synthesis system is realized in two stages: first, speech emotion recognition extracts soft emotion category likelihoods for the utterances; then a soft fusion of the emotion-dependent A2V mapping outputs form the affective facial synthesis. Experimental evaluations are performed on the SAVEE audio-visual dataset. The proposed models are assessed with objective and subjective evaluations. The proposed affective A2V system achieves significant MSE loss improvements in comparison to the recent literature. Furthermore, the resulting facial animations of the proposed system are preferred over the baseline animations in the subjective evaluations., Scientific and Technological Council of Turkey (TÜBİTAK)

We examine how experience-induced adaptations that affect the breadth of an ongoing activity affect performance. The research on organizational learning suggests that accumulating experience, both from repetition and adaptation at the activity level, improves outcomes. Yet, findings on the effects of increasing breadth—the number of different processes making up an activity—are mixed. Greater breadth exposes organizations to diverse activities. It also generates an additional need for coordination that may undermine performance. We examine the joint effect of experience and breadth on waste reduction for U.S. manufacturing facilities managing their toxic waste from 1991 to 2014. These facilities manage toxic waste on a chemical by chemical basis. We find a detrimental effect of breadth on performance that is highest for facilities with low experience; however, this effect is moderated by experience with the waste management activity. Because most facilities manage toxic waste from several chemicals, we also see spillovers—in terms of both learning benefits and the costs of increasing breadth. When a facility expands waste management breadth anywhere, performance decreases for the focal chemical. Yet, this spillover effect of breadth decreases for activities where the facility has accrued more experience. Our research clarifies when facilities should consider adding breadth to a routine activity and why performance in the proximate period may falter as the organization learns and improves in the longer term.

A pneumatic artificial muscle (PAM) based on a metal hydride (MH) is considered for a compact compliant actuator. It is suitable for board applications of human robot interaction (HRI). To address the problem of HRI representing by varying environment, a compliant control is introduced. In fact, the bottlenecks of improving the performance in the compliant control of the PAM actuator are: a) an inherent non-linear dynamics of a PAM, b) the parametric and non-linear uncertainties, influenced by varying environment, and c) an additional high dimension introduced by an MH employed as a driving force for the PAM. We propose a learning-based adaptive robust control (LARC) framework to tackle these challenges. A Bayesian learning technique deals with the parameter adaptation for the adaptive control. The effectiveness of the LARC has been examined in extensive experiments of tracking control.

Benefiting from the pay-as-you-go business model, cloud computing has significantly promoted service computing techniques in real-world industrial applications. Software applications based on cloud computing are becoming more and more popular. By integrating existing component cloud services through the internet, composite cloud systems can be built to meet sophisticated application logic. Stable execution of such systems is desirable in the long term so that the service-level agreements (SLAs), as well as users’ quality of experience (QoE), can be fulfilled. To achieve this goal, it is critical to identify and fault-tolerate system components at high risks of failing. This is extremely challenging due to the dynamic and uncertainty of the cloud environment that hosts the component cloud services. Nevertheless, existing approaches pay little attention to the modeling and analysis of system components’ reliability time series. To address the above issues, we first present a reliability evaluation method for component cloud services based on the reliability model and their failure probability under continuous client-side invocation tests. Then, we propose a perturbation-aware reliability sensitivity measurement approach (named PARS) for measuring the reliability sensitivity of component cloud services. It first analyzes the negative perturbations in component cloud services’ historical reliability time series based on the Markov chain rule. Then, it calculates the reliability sensitivity of component cloud services by analyzing how their reliability perturbations impact the reliability of the entire cloud system. To guarantee the execution quality of the composite cloud system, we further propose a proactive adaptation approach named PA-PARS that enables 1-out-of-2 N-version Programming fault-tolerance for composite cloud systems based on PARS. PA-PARS takes the reliability sensitivity of component cloud services estimated by PARS as input to assure the reliability of the cloud system. It consists of four parts: 1) risky system component identification; 2) adaptation trigger; 3) candidate component cloud service selection; and 4) NVP-based system construction as the proactive adaptation for the composite cloud system. The results of experiments conducted on two widely-used datasets demonstrate the effectiveness and efficiency of the proposed approaches in ensuring the reliability of composite cloud systems.

Modeling interactions among vehicles is critical in improving the efficiency and safety of autonomous driving since complex interactions are ubiquitous in many traffic scenarios. To model interactions under different traffic scenarios, most existing works consider interaction information implicitly in their specific tasks with hand-crafted features and pre-defined maneuvers. Extracting interaction representation which can be commonly used among different downstream tasks is not explored. In this paper, we propose a general and novel Graph Self-Attention Network (GSAN) to learn the spatial-temporal interaction representation among vehicles by a framework consisting of pre-training and fine-tuning. Specifically, in the pre-training step, we construct the GSAN module based on a graph self-attention layer and a gated recurrent unit layer and use trajectory auto-regression to learn the interaction information among vehicles. In the fine-tuning step, we propose two different adaptation schemes to utilize the learned interaction information in various downstream tasks and fine-tune the entire model with only a few steps. To illustrate the effectiveness and generality of our spatial-temporal interaction model, we conduct extensive experiments on two typical interaction-related tasks, namely, lane-changing classification and trajectory prediction. The experiment results demonstrate that our approach significantly outperforms the state-of-the-art solutions of these two tasks. We also visualize the impact of surrounding vehicles on the ego vehicle in different interaction scenes. The visualization offers an intuitive explanation on how our model captures the dynamic changing interactions among vehicles and makes good predictions in various interaction-related tasks.

This reflection article examines the importance of adapting cultural material to meet student needs and program objectives in two different educational contexts: Indonesia and Colombia. A needs analysis was done at both universities, resulting in material adaptation and development. While intercultural understanding and communication was the content of classes taught in both settings, the needs of students varied greatly. Adapting material for critical thinking (Üstünlüoglu, 2004) and gender roles were the focus in Indonesia; conversely, in Colombia combining TOEFL-prep within a content-based course (Coté, 2013) and incorporating diverse religious perspectives were needed. The necessity of considering specific student populations as well as program goals is discussed for future teaching implications.

Artificial intelligence (AI) has captured substantial interest from a wide array of marketing scholars in recent years. Our research contributes to this emerging domain by examining AI technologies in marketing via a global lens. Specifically, our lens focuses on three levels of analysis: country, company, and consumer. Our country-level analysis emphasizes the heterogeneity in economic inequality across countries due to the considerable economic resources necessary for AI adoption. Our company-level analysis focuses on glocalization because while the hardware that underlies these technologies may be global in nature, their application necessitates adaptation to local cultures. Our consumer-level analysis examines consumer ethics and privacy concerns, as AI technologies often collect, store and process a cornucopia of personal data across our globe. Through the prism of these three lenses, we focus on two important dimensions of AI technologies in marketing: (1) human-machine interaction and (2) automated analysis of text, audio, images, and video. We then explore the interaction between these two key dimensions of AI across our three-part global lens to develop a set of research questions for future marketing scholarship in this increasingly important domain.

The covariance matrix self-adaptation evolution strategy with repelling subpopulations (RS-CMSA-ES) is one of the most successful multimodal optimization methods currently available. However, some of its components may become inefficient in certain situations. This study introduces the second variant of this method, called RS-CMSA-ESII. It improves the adaptation schemes for the normalized taboo distances of the archived solutions and the covariance matrix of the subpopulation, the termination criteria for the subpopulations, and the way in which the infeasible solutions are treated. It also improves the time complexity of RS-CMSA-ES by updating the initialization procedure of a subpopulation and developing a more accurate metric for determining critical taboo regions. The effects of these modifications are illustrated by designing controlled numerical simulations. RS-CMSA-ESII is then compared with the most successful and recent niching methods for multimodal optimization on a widely adopted test suite. The results obtained reveal the superiority of RS-CMSA-ESII over these methods, including the winners of the competition on niching methods for multimodal optimization in previous years. Besides, this study extends RS-CMSA-ESII to dynamic multimodal optimization and compares it with a few recently proposed methods on the modified moving peak benchmark functions.

Deep learning (DL) theory has made great progress in the field of intelligent fault diagnosis, and the development of domain adaptation has greatly promoted fault diagnosis under polytropic working conditions (PWC). Extensive studies have been conducted to solve the problem of fault diagnosis under PWC. However, the existing fault diagnosis methods based on domain adaptation have the following shortcomings. First, multi-source information fusion is rarely considered. Second, the utilization of inherent labels is also insufficient in classification problems. To deal with the above problem, a novel multi-source dense adaptation adversarial network (MDAAN) is proposed, which leverages multi-sensor vibration information and classification label information. Specifically, the frequency spectrum of multi-sensor data are firstly employed to make full use of fault information. Afterwards, the dense convolution and fusion convolution blocks are used for deep feature extraction and fusion. Finally, a joint loss function is reconstructed under the framework of unsupervised learning, which considers the distribution differences of the features and the label information simultaneously. The experimental results from various working conditions, including still distant working conditions, all demonstrate that the proposed method can achieve state-of-the-art performances, which has shown great promise as an intelligent fault diagnosis method.

Nuclear fusion is a promising alternative to address the problem of sustainable energy production. The tokamak is an approach to fusion based on magnetic plasma confinement, constituting a complex physical system with many control challenges. We study the characteristics and optimization of reservoir computing (RC) for real-time and adaptive prediction of plasma profiles in the DIII-D tokamak. Our experiments demonstrate that RC achieves comparable results to state-of-the-art (deep) convolutional neural networks (CNNs) and long short-term memory (LSTM) models, with a significantly easier and faster training procedure. This efficient approach allows for fast and frequent adaptation of the model to new situations, such as changing plasma conditions or different fusion devices.