Hasil untuk "Cybernetics"

Manuel Baltieri, Martin Biehl, Matteo Capucci et al.

The internal model principle, originally proposed in the theory of control of linear systems, nowadays represents a more general class of results in control theory and cybernetics. The central claim of these results is that, under suitable assumptions, if a system (a controller) can regulate against a class of external inputs (from the environment), it is because the system contains a model of the system causing these inputs, which can be used to generate signals counteracting them. Similar claims on the role of internal models appear also in cognitive science, especially in modern Bayesian treatments of cognitive agents, often suggesting that a system (a human subject, or some other agent) models its environment to adapt against disturbances and perform goal-directed behaviour. It is however unclear whether the Bayesian internal models discussed in cognitive science bear any formal relation to the internal models invoked in standard treatments of control theory. Here, we first review the internal model principle and present a precise formulation of it using concepts inspired by categorical systems theory. This leads to a formal definition of ``model'' generalising its use in the internal model principle. Although this notion of model is not a priori related to the notion of Bayesian reasoning, we show that it can be seen as a special case of possibilistic Bayesian filtering. This result is based on a recent line of work formalising, using Markov categories, a notion of ``interpretation'', describing when a system can be interpreted as performing Bayesian filtering on an outside world in a consistent way.

Xuefei Wu, Xiao Yin, Yuanyang Zhu et al.

Efficient exploration in multi-agent reinforcement learning (MARL) is a challenging problem when receiving only a team reward, especially in environments with sparse rewards. A powerful method to mitigate this issue involves crafting dense individual rewards to guide the agents toward efficient exploration. However, individual rewards generally rely on manually engineered shaping-reward functions that lack high-order intelligence, thus it behaves ineffectively than humans regarding learning and generalization in complex problems. To tackle these issues, we combine the above two paradigms and propose a novel framework, LIGHT (Learning Individual Intrinsic reward via Incorporating Generalized Human experTise), which can integrate human knowledge into MARL algorithms in an end-to-end manner. LIGHT guides each agent to avoid unnecessary exploration by considering both individual action distribution and human expertise preference distribution. Then, LIGHT designs individual intrinsic rewards for each agent based on actionable representational transformation relevant to Q-learning so that the agents align their action preferences with the human expertise while maximizing the joint action value. Experimental results demonstrate the superiority of our method over representative baselines regarding performance and better knowledge reusability across different sparse-reward tasks on challenging scenarios.

Erik Derner, Kristina Batistic, Jan Zahalka et al.

As large language models (LLMs) permeate more and more applications, an assessment of their associated security risks becomes increasingly necessary. The potential for exploitation by malicious actors, ranging from disinformation to data breaches and reputation damage, is substantial. This paper addresses a gap in current research by specifically focusing on security risks posed by LLMs within the prompt-based interaction scheme, which extends beyond the widely covered ethical and societal implications. Our work proposes a taxonomy of security risks along the user-model communication pipeline and categorizes the attacks by target and attack type alongside the commonly used confidentiality, integrity, and availability (CIA) triad. The taxonomy is reinforced with specific attack examples to showcase the real-world impact of these risks. Through this taxonomy, we aim to inform the development of robust and secure LLM applications, enhancing their safety and trustworthiness.

L.A. Gorodetskaya, A.Y. Denisova, L.M. Kavelenova et al.

Rare plant species restoration (reintroduction) is one of the main biodiversity conservation activities. Reintroduced plants need constant monitoring in order to study features of their development and control the population state. To reduce the human impact on the natural habitat of plants and simplify the monitoring process, we propose the use of automatic analysis of unmanned aerial vehicles (UAVs) data using the Yolov3 neural network. The article discusses neural network parameters for detecting Paeonia Tenuifolia, reintroduced in the Samara region by ecologists from the Department of Ecology, Botany and Nature Conservation of Samara University. The main issue under research is the possibility of training a neural network from peony images collected in an artificial habitat with a subsequent application to images collected in a natural habitat and the possibilities of using multi-temporal data to improve the network training quality. The experiments have shown that training a neural network exclusively using images collected in the natural habitat makes it possible to achieve a probability of correct detection of peonies of 0.93, while using data obtained at different years allows increasing the probability of correct detection to 0.95.

Chi Zhang

In this paper, a system for automatic word segmentation of Chinese criminal judgment documents is proposed. The system uses a hybrid model composed of fine-tuned BERT (Bidirectional Encoder Representations from Transformers), BiLSTM (Bidirectional Long Short Term Memory) and CRF (Conditional Random Field) for named entity recognition, and introduces a custom dictionary that includes common professional terms in Chinese criminal trial documents, as well as a rule system based on judicial system and litigation procedure related regulations, to further improve the accuracy of word segmentation. BERT uses a deep bidirectional Transformer encoder to pre-train general language representations from large-scale unlabeled text corpora. BiLSTM uses two LSTM networks, one for the forward direction and one for the backward direction, to capture the context from both sides of the input sequence. CRF uses a set of features and weights to define a log-linear distribution over the output sequence. Experimental results show that the proposed system has significantly improved word segmentation accuracy compared to the current commonly used Chinese word segmentation models. In the results of the segmentation of the test data, the F1 scores for jieba, THULAC and the segmentation system proposed in this paper are 85.59%, 87.94% and 94.82%, respectively.

Muhsen Dena Kadhim, Raheem Firas Abdulrazzaq, Sadiq Ahmed T.

Recent advances in path-planning algorithms have transformed robotics. The Rapidly exploring Random Tree (RRT) algorithm underpins autonomous robot navigation. This paper systematically examines the uses and development of RRT algorithms in single and multiple robots to demonstrate their importance in modern robotics studies. To do this, we have reviewed 70 works on RRT algorithms in single and multiple robot path planning from 2015 to 2023. RRT algorithm evolution, including crucial turning points and innovative techniques, have been examined. A detailed comparison of the RRT Algorithm versions reveals their merits, limitations, and development potential. The review’s identification of developing regions and future research initiatives will enable roboticists to use RRT algorithms. This thorough review is essential to the robotics community, inspiring new ideas, helping problem-solving, and expediting single- and multi-robot system development. This highlights the necessity of RRT algorithms for autonomous and collaborative robotics advancement.

Xuanting Xie, Erlin Pan, Zhao Kang et al.

Graph clustering, an important unsupervised problem, has been shown to be more resistant to advances in Graph Neural Networks (GNNs). In addition, almost all clustering methods focus on homophilic graphs and ignore heterophily. This significantly limits their applicability in practice, since real-world graphs exhibit a structural disparity and cannot simply be classified as homophily and heterophily. Thus, a principled way to handle practical graphs is urgently needed. To fill this gap, we provide a novel solution with theoretical support. Interestingly, we find that most homophilic and heterophilic edges can be correctly identified on the basis of neighbor information. Motivated by this finding, we construct two graphs that are highly homophilic and heterophilic, respectively. They are used to build low-pass and high-pass filters to capture holistic information. Important features are further enhanced by the squeeze-and-excitation block. We validate our approach through extensive experiments on both homophilic and heterophilic graphs. Empirical results demonstrate the superiority of our method compared to state-of-the-art clustering methods.

Yan Zhang, Ming Jia, Meng Li et al.

Operators experience complicated physiological and psychological states when exposed to extreme heat stress, which can impair cognitive function and decrease performance significantly, ultimately leading to severe secondary disasters. Therefore, there is an urgent need for a feasible technique to identify their abnormal states to enhance the reliability of human-cybernetics systems. With the advancement of deep learning in physiological modeling, a model for evaluating operators' performance driven by electrocardiogram (ECG) and functional near-infrared spectroscopy (fNIRS) was proposed, demonstrating high ecological validity. The model fused a convolutional neural network (CNN) backbone and a graph attention network (GAT) backbone to extract discriminative features from ECG time-frequency spectrums and fNIRS prefrontal cortex (PFC) network respectively with deeper neuroscience domain knowledge, and eventually achieved 0.90 AUC. Results supported that handcrafted features extracted by specialized neuroscience methods can alleviate overfitting. Inspired by the small-world nature of the brain network, the fNIRS PFC network was organized as an undirected graph and embedded by GAT. It is proven to perform better in information aggregation and delivery compared to a simple non-linear transformation. The model provides a potential neuroergonomics application for evaluating the human state in vital human-cybernetics systems under industry 5.0 scenarios.

Issatay Tokmurziyev, Valerii Serpiva, Alexey Fedoseev et al.

This paper presents GazeRace, a novel system that leverages eye-tracking technology for intuitive drone control. Using the MediaPipe library, the system translates eye movements into precise drone commands, enabling effective remote piloting. In testing, GazeRace demonstrated an 18% reduction in drone trajectory length while maintaining competitive speed with traditional controls. The results suggest that this approach enhances control accuracy and reduces user frustration, offering a significant advancement in the field of human-computer interaction and drone navigation.

Zhonghua Wang

Providing personalized recommendation service for users and improving the accuracy of recommendation and user satisfaction are the main research tasks of current travel recommendation systems. The intelligent recommendation model of tourist places requires the algorithm to be able to accurately recommend tourist attractions according to the user’s interests. Big data and artificial intelligence technologies have driven the development of intelligent recommendation systems. However, realistic data are often sparse, and the lack of common rating items among users makes some traditional similarity measures impossible to calculate. In addition, traditional collaborative filtering algorithms ignore the issue of user preferences, which can cause a decrease in recommendation accuracy. To address these issues, this paper analyzes the factors affecting users’ interest preferences in terms of their global and local rating information. The interest preferences of users are calculated by computing the probability distribution of user rating information globally and using Hamming approach degree. A similarity algorithm about user preferences is derived using Jeffries-Matusita distance. The similarity algorithm is effectively combined with the traditional similarity algorithm to propose a model of collaborative filtering recommendation algorithm for tourist attractions based on user preferences under sparse data. The paper aims to improve the accuracy of tourist attraction recommendation systems by incorporating user preferences and addressing the issue of sparse data and the lack of common rating items among users that traditional similarity measures cannot calculate. The experimental results show that the improved algorithm model outperforms the traditional collaborative filtering algorithm and other algorithms. And it also has high accuracy rate on more sparse tourism data set.

Songsong Ma, Yuansheng Chen, Songlin Yang et al.

Tens of crawling bio-robots with cockroaches as the mobile platform have been developed with various functions. Compared with artificial crawling robots of the same size, they revealed better flexibility, larger payload, and stronger endurance. These features made bio-robots ideal for pipeline inspection scenarios because the advancements in locomotion mechanisms and efficient power systems are still hurdles for current artificial systems. In this study, we controlled the bio-robot to crawl in the confined dark pipeline and achieved autonomous motion control with the help of an onboard sensing system. Specifically, a micro-camera was mounted on the electronic backpack of the cockroach for image collection, and an IMU sensor was used to compute its body orientation. The electronic backpack transmitted images to the host computer for junction recognition and distance estimation. Meanwhile, the insect's habituation to electrical stimulation has long been an uncertain factor in the control of bio-robots. Here, a synergistic stimulation strategy was proposed to markedly reduce the habituation and increase the number of effective turning controls to over 100 times. It is also found that both the increase of payload and the application of stimulations could promote the metabolic rate by monitoring carbon dioxide release. With the integration of synergistic stimulation and autonomous control, we demonstrated the fully autonomous pipeline navigation with our cockroach bio-robot, which realized the cycle number of approximately 10 in a roll. This research provides a novel technology that has the potential for practical applications in the future.

Alexander Petrovsky, Yomna Youssef, Kirill Myasoedov et al.

This paper focuses on Passable Obstacles Aware (POA) planner - a novel navigation method for two-wheeled robots in a highly cluttered environment. The navigation algorithm detects and classifies objects to distinguish two types of obstacles - passable and unpassable. Our algorithm allows two-wheeled robots to find a path through passable obstacles. Such a solution helps the robot working in areas inaccessible to standard path planners and find optimal trajectories in scenarios with a high number of objects in the robot's vicinity. The POA planner can be embedded into other planning algorithms and enables them to build a path through obstacles. Our method decreases path length and the total travel time to the final destination up to 43% and 39%, respectively, comparing to standard path planners such as GVD, A*, and RRT*

Oleksandr Kolgatin, Dmytro Holubnychyi, Larysa Kolgatina

The research investigates the systematicity of students' learning activity as a parameter of the student's model, as well as the influence of systematicity on learning outcomes in the course "Operating Systems." As a theoretical framework, the importance of equipping the student himself as the subject of the educational process with the abilities and proper pedagogical forecasting tools for independent choice of the right variation of educational activity is demonstrated. Model parameters in such a pedagogical diagnostics system are proposed and analyzed. Empirical work has been completed on the basis of the learning management system Moodle and allows for the analysis of the association between students' timeliness in completing learning tasks and their educational achievements, as well as the structure of students' time planning at homework. Recommendations to improve the educational process have been suggested.

Tyng-Ruey Chuang, Rebecca C. Fan, Ming-Syuan Ho et al.

The nature and extent of openness depend on the context and/or disciplinary domain. Earlier usage of the term open was in the context of computer systems. For example, in networked systems of computers, ‘openness’ refers to enabling protocols that connect previously closed systems so that they can communicate with each other. Beyond that, openness has been used to imply a spectrum of meanings, notably since the campaign for open source software development populated the term ‘open’ and its suggested notions of ‘openness’ as freedom, entitlement, or norm. As a social form of organising, ‘openness’ suggests a way of sharing resources. In the corporate context, ‘openness’ refers to more active involvement of stakeholders in the process of value creation.

Euler Evangelista, Cristiana De Muÿlder

This study presents a proposal to build and analyze a domain ontology as a tool to support the knowledge transfer process in the context of software requirements analysis in the medical/pharmaceutical industry. The proposal is to use ontologies as an engineering artifact with the objective of representing knowledge in a specific domain, which, in the context of this research, is software modeling. A domain ontology is built to represent the requirements of a data warehouse/business intelligence software in the medical/pharmaceutical industry. The ontology-building process is supported by a specific methodology, defined with the purpose of building such artifacts, named "Methondology," and selected based on the research requirements. A prototype is created in the implementation phase of the ontology-building process. The results demonstrate that ontology domains can contribute to the process of analyzing and representing software requirements, as well as serving as a tool for organizational knowledge transfer through continuous knowledge conversion, which is critical for business sustainability. This study is an attempt to understand the knowledge conversion process in software development projects. Tacit knowledge is complex to articulate through formal language once it has been embedded with individual experience.

Jun Wan, Hui Xi, Jie Zhou et al.

Current fully-supervised facial landmark detection methods have progressed rapidly and achieved remarkable performance. However, they still suffer when coping with faces under large poses and heavy occlusions for inaccurate facial shape constraints and insufficient labeled training samples. In this paper, we propose a semi-supervised framework, i.e., a Self-Calibrated Pose Attention Network (SCPAN) to achieve more robust and precise facial landmark detection in challenging scenarios. To be specific, a Boundary-Aware Landmark Intensity (BALI) field is proposed to model more effective facial shape constraints by fusing boundary and landmark intensity field information. Moreover, a Self-Calibrated Pose Attention (SCPA) model is designed to provide a self-learned objective function that enforces intermediate supervision without label information by introducing a self-calibrated mechanism and a pose attention mask. We show that by integrating the BALI fields and SCPA model into a novel self-calibrated pose attention network, more facial prior knowledge can be learned and the detection accuracy and robustness of our method for faces with large poses and heavy occlusions have been improved. The experimental results obtained for challenging benchmark datasets demonstrate that our approach outperforms state-of-the-art methods in the literature.

Bin Sun, Shaofan Wang, Dehui Kong et al.

3D action recognition is referred to as the classification of action sequences which consist of 3D skeleton joints. While many research work are devoted to 3D action recognition, it mainly suffers from three problems: highly complicated articulation, a great amount of noise, and a low implementation efficiency. To tackle all these problems, we propose a real-time 3D action recognition framework by integrating the locally aggregated kinematic-guided skeletonlet (LAKS) with a supervised hashing-by-analysis (SHA) model. We first define the skeletonlet as a few combinations of joint offsets grouped in terms of kinematic principle, and then represent an action sequence using LAKS, which consists of a denoising phase and a locally aggregating phase. The denoising phase detects the noisy action data and adjust it by replacing all the features within it with the features of the corresponding previous frame, while the locally aggregating phase sums the difference between an offset feature of the skeletonlet and its cluster center together over all the offset features of the sequence. Finally, the SHA model which combines sparse representation with a hashing model, aiming at promoting the recognition accuracy while maintaining a high efficiency. Experimental results on MSRAction3D, UTKinectAction3D and Florence3DAction datasets demonstrate that the proposed method outperforms state-of-the-art methods in both recognition accuracy and implementation efficiency.

Shaoxiong Ji, Yue Tan, Teemu Saravirta et al.

Federated learning is a new learning paradigm that decouples data collection and model training via multi-party computation and model aggregation. As a flexible learning setting, federated learning has the potential to integrate with other learning frameworks. We conduct a focused survey of federated learning in conjunction with other learning algorithms. Specifically, we explore various learning algorithms to improve the vanilla federated averaging algorithm and review model fusion methods such as adaptive aggregation, regularization, clustered methods, and Bayesian methods. Following the emerging trends, we also discuss federated learning in the intersection with other learning paradigms, termed federated X learning, where X includes multitask learning, meta-learning, transfer learning, unsupervised learning, and reinforcement learning. In addition to reviewing state-of-the-art studies, this paper also identifies key challenges and applications in this field, while also highlighting promising future directions.

Elena Tverytnykova, Tatyana Drozdova, Yulia Demidova

The results of activity and stages of implementation of various measures of environmental policy of the enterprise are considered on the example of Shebelinsky department of gas condensate and oil processing. The structural scheme of the ecological management system model as an important component of the integrated management system of the gas condensate and oil refining enterprise is proposed on the basis of scientific literature, regulatory and legal documents using structural-logical method, systematization and generalization, and methods of meaningful and comparative analysis. This work also describes the specifics of ecological standards implementation based on ISO 9000 and ISO 14000 series for integrated management system development. It has been proved in practice that development and implementation of the integrated management system in the natural gas liquids and oil processing enterprise provides certain advantages, specifically, improves the overall products competitiveness, helps to adapt faster to the market environment, shapes environmental policy and approaches for effective resource management. The integrated management system development process consists of 4 basic stages, namely, planning, implementation, verification, and adjustment. Complex process and system approach represents an integral part of the integrated management system development. While being gradually implemented, the application of the additive and simultaneous integration models made it possible to create a single management system based on the requirements of different standards and significantly reduce the documentation volume.

Serhii Chalyi, Volodymyr Leshchynskyi, Iryna Leshchynska

The subject matter of the article is the processes of automated construction of explanations on the operation of an intelligent system for use in the human-computer interface. The goal is to develop a dynamic model of explanation for the human-computer interface using temporal knowledge about the process of functioning of the intelligent system. Temporal knowledge makes it possible to set possible sequences of decision-making actions in an intelligent system based on the known temporal order for pairs of such actions. Tasks: to develop an approach to constructing explanations for the operation of an intelligent system based on the use of temporal knowledge; development of a three-aspect model of explanations using temporal knowledge. The approaches used are: approaches to the construction of knowledge representation based on temporal dependencies, approaches to the construction of chatbot answers using rules, as well as with their automatic generation. The following results are obtained. The structuring of aspects of explanation taking into account the possibilities of their description with the help of temporal knowledge is performed; a temporal approach to constructing an explanation is proposed; a dynamic explanation model using temporal rules has been developed. Conclusions. The scientific novelty of the results is as follows. A temporal approach to constructing explanations for the operation of an intelligent system is proposed. The approach describes explanation as a process consisting of a temporally ordered sequence of facts. The order of time for pairs of facts is determined by temporal rules. Such rules may define the explanation process with varying degrees of detail over time, depending on the request for clarification. Detailed explanations reflect the subject area model and include the basic and alternative sequences of actions performed by the intelligent system. The explanation of the basic patterns of the intelligent system makes it possible to interpret the limitations that affect the obtained solution. The explanation of the system as a whole provides an implicit reflection of the key causal relationships, which allows you to get a simplified interpretation of the results of the intelligent system. A dynamic model of describing explanations based on temporal knowledge for use in the human-computer interface is proposed. The model takes into account the description of actions in the subject area, the patterns of these actions, as well as generalized causal relationships between such patterns. The model provides an opportunity to present the dynamics of the process of functioning of the intelligent system with the required level of detail, as well as change the level of detail to clarify the explanation at the request of the user.