Learning Mechanism Research Articles

Research on User Data Protection for Electric-driven Bicycle

Under the fast growth of new industries such as the Internet of Things, cloud computing, and artificial intelligence, the shared electric-driven bicycle has proved its great market competitiveness as a product of the era of mobile internet and sharing economy. However, the frequent interaction of large-scale data and the cross-fertilization of massively stored information have exposed a series of problems such as user privacy leakage, data island, imperfect credit systems, etc. For this objective, a data security protection model for shared electric-driven bicycles has been developed. To replace the centralized update mechanism and Blockchain technology, the model uses SM2 and Blockchain identity authentication protocols, decentralized cryptographic storage and multi-factor authentication, asynchronous federated learning mechanism, and distributed end-to-end parameter update mechanism. The experimental results show that the adoption of this data protection model can effectively resist replay attacks and prevent attackers from decrypting user identity information offline. It can also mitigate the risk of parameter leakage caused by centralized servers and establish a secure and trustworthy distributed data sharing mechanism for users of the shared electric-driven bicycle network.

A Novel Evolutionary Algorithm With Column and Sub-Block Local Search for Sudoku Puzzles

Sudoku puzzles are not only popular intellectual games but also NP-hard combinatorial problems related to various real-world applications, which have attracted much attention worldwide. Although many efficient tools, such as evolutionary computation (EC) algorithms, have been proposed for solving Sudoku puzzles, they still face great challenges with regard to hard and large instances of Sudoku puzzles. Therefore, to efficiently solve Sudoku puzzles, this paper proposes a genetic algorithm (GA)-based method with a novel local search technology called local search-based GA (LSGA). The LSGA includes three novel design aspects. First, it adopts a matrix coding scheme to represent individuals and designs the corresponding crossover and mutation operations. Second, a novel local search strategy based on column search and sub-block search is proposed to increase the convergence speed of the GA. Third, an elite population learning mechanism is proposed to let the population evolve by learning the historical optimal solution. Based on the above technologies, LSGA can greatly improve the search ability for solving complex Sudoku puzzles. LSGA is compared with some state-of-the-art algorithms at Sudoku puzzles of different difficulty levels and the results show that LSGA performs well in terms of both convergence speed and success rates on the tested Sudoku puzzle instances.

A multi-surrogate genetic programming hyper-heuristic algorithm for the manufacturing project scheduling problem with setup times under dynamic and interference environments

In this study, in order to cope with the uncertain environments and space interference scenarios encountered in the production process for a class of manufacturing projects, we propose a novel manufacturing project scheduling problem with setup times under dynamic and interference environments (MPSPST-DIE), and design a novel multi-surrogate genetic programming hyper-heuristic (HH-MGP) algorithm to address it. Firstly, MPSPST-DIE is required to make decisions on the activity schedule, resource setup and space allocation. Therefore, we modify the traditional resource based policy class that only contains the activity schedule and simulate the entire scheduling process. Secondly, a new hyper-heuristic genetic programming algorithm is designed to automatically evolve activity rules, resource setup rules and space allocation rules simultaneously. Moreover, the multi-surrogate is devised to improve the performance of the basic genetic programming (GP) algorithm. In addition, a new evolutionary learning mechanism is embedded in the multi-surrogate. Different surrogates learn from each other to complement each other’s strengths. Finally, numerical instances of the MPSPST-DIE are generated by configuring specific parameters of spatial resources and extensive numerical experiments are performed. At the same time, we implement the Taguchi experiment for the sensitivity analysis of parameters. The comparative analysis between the HH-MGP and traditional rules is performed. Further, the performance comparison between the multi-surrogate and other surrogates is also conducted. The experimental results show that the evolved rule of the HH-MGP performs better than the traditional rules for the MPSPST-DIE. The performance of the multi-surrogate model added to the GP algorithm are generally better than the single-surrogate model and no-surrogate.

Multi-channel high-order network representation learning research.

The existing network representation learning algorithms mainly model the relationship between network nodes based on the structural features of the network, or use text features, hierarchical features and other external attributes to realize the network joint representation learning. Capturing global features of the network allows the obtained node vectors to retain more comprehensive feature information during training, thereby enhancing the quality of embeddings. In order to preserve the global structural features of the network in the training results, we employed a multi-channel learning approach to perform high-order feature modeling on the network. We proposed a novel algorithm for multi-channel high-order network representation learning, referred to as the Multi-Channel High-Order Network Representation (MHNR) algorithm. This algorithm initially constructs high-order network features from the original network structure, thereby transforming the single-channel network representation learning process into a multi-channel high-order network representation learning process. Then, for each single-channel network representation learning process, the novel graph assimilation mechanism is introduced in the algorithm, so as to realize the high-order network structure modeling mechanism in the single-channel network representation learning. Finally, the algorithm integrates the multi-channel and single-channel mechanism of high-order network structure joint modeling, realizing the efficient use of network structure features and sufficient modeling. Experimental results show that the node classification performance of the proposed MHNR algorithm reaches a good order on Citeseer, Cora, and DBLP data, and its node classification performance is better than that of the comparison algorithm used in this paper. In addition, when the vector length is optimized, the average classification accuracy of nodes of the proposed algorithm is up to 12.24% higher than that of the DeepWalk algorithm. Therefore, the node classification performance of the proposed algorithm can reach the current optimal order only based on the structural features of the network under the condition of no external feature supplementary modeling.

Learning from Failures: Differentiating Between Slip-ups and Knowledge Gaps

Our research investigates firm learning from failures by dividing them into two types, failures that occur due to slip-ups and those that occur due to knowledge gaps, and by examining whether learning occurs in the context of both types of failures. We study these phenomena in the context of product recalls in pharmaceuticals and medical devices. Based on text analysis of recall documents, recalls are divided into process related and design related to represent slip-up failures and knowledge gap failures. We further study how innovation capabilities, represented by accumulated stocks of patents and lagged research and development (R&D) intensity, impact learning from both types of failures. We test our hypotheses using negative binomial generalized linear models to analyze longitudinal data for 108 publicly traded U.S. firms over 2000–2016 comprising 7,984 recalls. Results indicate that design-related recalls generate learning to a greater extent than process-related recalls, and that accumulated patents and lagged R&D intensity enhance learning from design-related recalls. These findings suggest that the learning mechanisms invoked by failures are concentrated more on knowledge gap failures than slip-up failures, and such learning is impacted by innovation capabilities. Overall, this research extends organizational learning theory by differentiating between learning from different types of failures and extends absorptive capacity theory by incorporating the role of innovation capabilities in enhancing learning from failures. We develop recommendations for learning from slip-up failures by focusing on the cultural and social mechanisms of organizational learning in addition to the technical and structural mechanisms that may mainly impact learning from knowledge gap failures. Supplemental Material: The online appendices are available at https://doi.org/10.1287/orsc.2021.15663 .

Dopamine Receptor-Expressing Neurons Are Differently Distributed throughout Layers of the Motor Cortex to Control Dexterity.

The motor cortex comprises the primary descending circuits for flexible control of voluntary movements and is critically involved in motor skill learning. Motor skill learning is impaired in patients with Parkinson's disease, but the precise mechanisms of motor control and skill learning are still not well understood. Here we have used transgenic mice, electrophysiology, in situ hybridization, and neural tract-tracing methods to target genetically defined cell types expressing D1 and D2 dopamine receptors in the motor cortex. We observed that putative D1 and D2 dopamine receptor-expressing neurons (D1+ and D2+, respectively) are organized in highly segregated, nonoverlapping populations. Moreover, based on ex vivo patch-clamp recordings, we showed that D1+ and D2+ cells have distinct morphological and electrophysiological properties. Finally, we observed that chemogenetic inhibition of D2+, but not D1+, neurons disrupts skilled forelimb reaching in adult mice. Overall, these results demonstrate that dopamine receptor-expressing cells in the motor cortex are highly segregated and play a specialized role in manual dexterity.

Virtual Communities of Practice on LinkedIn: A Study on Self-directed, Work-related Learning

As a career network, LinkedIn offers access to a wide range of job-related information and a low-threshold opportunity for exchange among experts in various fields. The purpose of this study was to find out which LinkedIn functions are particularly useful for self-directed professional development and to what extent users’ motivation to learn is influenced by their virtual community of practice (VCoP) on LinkedIn. Two approaches were taken to answer this question. First, a literature review on four different learning theories was conducted: self-directed learning, situated learning, social learning and connectivism. The learning mechanisms on social media were then put into the context of those four learning theories. Additionally, the specific potential of learning via social media and especially via LinkedIn was considered. Second, a qualitative interview study with LinkedIn users who were categorised as digital learning experts based on their education and/or profession was conducted. According to the results of the study, LinkedIn offers excellent conditions for professional development. Various functions such as consuming others’ posts, writing own posts, comments or private messages and subscribing to hashtags or newsletters support work-related learning processes. The value of the LinkedIn Learning course platform is controversial. The results also show that VCoP activities can influence users’ motivation to learn both positively and negatively. The LinkedIn activities of others can inspire users, but they can also cause social pressure. The results of the interview study represent an addition to the research field of “learning with social media” and can be used as recommendations for dealing with LinkedIn in order to broaden or deepen one’s professional knowledge. This study especially highlights the use of LinkedIn for lifelong learning with the goal of professional development.

Using Deep Learning for the Taxonomic Classification of Microbial Sequences

Microbes are common creatures and play a crucial role in our world. Thus, the understanding of microbial communities brings benefits to human lives. Because the material samples of microbes contain sequences belonging to different organisms, an important task in analyzing processes is to classify the sequences into groups of different species or closely related organisms, called metagenomic classification. Many classification approaches were proposed to analyze the metagenomic data. However, due to the complexity of microbial samples, the accuracy performance of those methods still remains a challenge. This study applies an effective deep learning framework for the classification of microbial sequences. The proposed architecture combines a sequence embedding layer with other layers of a bidirectional Long Short-Term Memory, Seft-attention, and Dropout mechanisms for feature learning. Experimental results demonstrate the strength of the proposed method on datasets of real metagenomes.

OBE 理念下计算机网络课程教学改革研究

Computer network is the core course of network engineering and related majors, which provides support for subsequent professional learning. In response to the pain points of traditional computer network teaching, such as the complexity of course theory and poor applicability, lack of incentive mechanisms for active learning, low self-efficacy of students in learning, and lack of process evaluation, this paper proposes a hybrid teaching approach based on the OBE education concept and a teaching model of “F-BOPPPS+M+N/”. The results of data comparison before and after the reform show that, The achievement level of this course has significantly improved in theoretical testing and experimental evaluation. This article introduces the reform path from the aspects of teaching objectives, teaching content, teaching resources, teaching methods, teaching modes, etc. Finally, the reform effect is introduced.

Forecasting agricultures security indices: Evidence from transformers method

AbstractIn recent years, ensuring food security has become a global concern, necessitating accurate forecasting of agriculture security to aid in policymaking and resource allocation. This article proposes the utilization of transformers, a powerful deep learning technique, for predicting the Agriculture Security Index ( ). The is a comprehensive metric that evaluates the stability and resilience of agricultural systems. By harnessing the temporal dependencies and complex patterns present in historical data, transformers offer a promising approach for accurate and reliable forecasting. The transformer architecture, renowned for its ability to capture long‐range dependencies, is tailored to suit the forecasting task. The model is trained using a combination of supervised learning and attention mechanisms to identify salient features and capture intricate relationships within the data. To evaluate the performance of the proposed method, various evaluation metrics, including mean absolute error, root mean square error, and coefficient of determination, are employed to assess the accuracy, robustness, and generalizability of the transformer‐based forecasting approach. The results obtained demonstrate the efficacy of transformers in forecasting the , outperforming traditional time series forecasting methods. The transformer model showcases its ability to capture both short‐term fluctuations and long‐term trends in the , allowing policymakers and stakeholders to make informed decisions. Additionally, the study identifies key factors that significantly influence agriculture security, providing valuable insights for proactive intervention and resource allocation.

Turbofan Engine Health Prediction Model Based on ESO-BP Neural Network

Predicting the health index of turbofan engines is critical in reducing downtime and ensuring aircraft safety. This study introduces the elite snake optimizer-back propagation (ESO-BP) model to address the challenges of low accuracy and poor stability in predicting the health index of turbofan engines through neural networks. Firstly, the snake optimizer (SO) was improved into the elite snake optimizer (ESO) through an elite-guided strategy and a reverse learning mechanism. The performance improvement was validated using benchmark functions. Additionally, feature importance was introduced as a feature selection method. Finally, the optimization results of the ESO were employed to set the initial weights and biases of the BP neural network, preventing convergence to local optima. The prediction performance of the ESO-BP model was validated using the C-MAPSS datasets. The ESO-BP model was compared with the CNN, RNN, LSTM, baseline BP, and unimproved SO-BP models. The results demonstrated that the ESO-BP model has a superior accuracy with an impressive R-squared (R2) value of 0.931 and a root mean square error (RMSE) of 0.060 on the FD001 sub-dataset. Furthermore, the ESO-BP model exhibited lower standard deviations of evaluation metrics on 100 trials. According to the study, ESO-BP demonstrated a greater prediction accuracy and stability when compared to commonly used models such as CNN, RNN, LSTM, and BP.

Adaptive generative AI for dynamic cybersecurity threat detection in enterprises

This research paper provides a thorough examination of the application of Generative Artificial Intelligence (AI) in the context of dynamic cybersecurity threat detection within enterprises. Recognizing the evolving nature of cyber threats, the study focuses on adaptive generative AI models designed to enhance threat detection capabilities. Through an extensive review of existing literature and case studies, the paper explores various Adaptive Generative AI methodologies, including machine learning algorithms, continuous learning mechanisms, and real-time data processing. The analysis encompasses the strengths and limitations of these approaches, shedding light on their efficacy in addressing the complex and dynamic cybersecurity landscape. By offering a comprehensive overview, this research aims to guide the development and implementation of adaptive generative AI solutions for effective threat detection and mitigation in enterprise cybersecurity.

Principle of Educational Management in Hybrid Learning

Hybrid learning occasioned by the introduction of the traditional face to face instruction with online learning mechanisms, has developed as a prominent educational approach in response to the developing landscape of teaching and process of learning. The professional management of hybrid learning ecosystems is important for ensuring best students’ engagement, the outcomes of learning, and the general success. The study investigates the principles of educational management of hybrid learning environments highlighting basic consideration and methods for the management. Based on the established principles of educational management, such as effective communication, support system, assessment and curriculum patterns, this study investigates how these principles apply in the realm of hybrid learning. In addition to that, the study explores the contribution of technology in helping management processes and improving student experiences in hybrid environments. Understanding and adopting the tenants of educational management in hybrid learning setups, educators can come up with dynamic, inclusive, and effective learning experiences that satisfies the diverse demands of the students and promote academic excellence

Improving the use of LLMs in radiology through prompt engineering: from precision prompts to zero-shot learning.

Large language models (LLMs) such as ChatGPT have shown significant potential in radiology. Their effectiveness often depends on prompt engineering, which optimizes the interaction with the chatbot for accurate results. Here, we highlight the critical role of prompt engineering in tailoring the LLMs' responses to specific medical tasks. Using a clinical case, we elucidate different prompting strategies to adapt the LLM ChatGPT using GPT4 to new tasks without additional training of the base model. These approaches range from precision prompts to advanced in-context methods such as few-shot and zero-shot learning. Additionally, the significance of embeddings, which serve as a data representation technique, is discussed. Prompt engineering substantially improved and focused the chatbot's output. Moreover, embedding of specialized knowledge allows for more transparent insight into the model's decision-making and thus enhances trust. Despite certain challenges, prompt engineering plays a pivotal role in harnessing the potential of LLMs for specialized tasks in the medical domain, particularly radiology. As LLMs continue to evolve, techniques like few-shot learning, zero-shot learning, and embedding-based retrieval mechanisms will become indispensable in delivering tailored outputs. · Large language models might impact radiological practice and decision-masking.. · However, implementation and performance are dependent on the assigned task.. · Optimization of prompting strategies can substantially improve model performance.. · Strategies for prompt engineering range from precision prompts to zero-shot learning..

Intelligent Situation Awareness Based on the Fractal Dimension and Multidimensional Reconstruction

AbstractIn order to realize autonomy and intelligence for situation awareness, this paper proposes an intelligent situation awareness model based on fractional dimension information mining and multidimensional information reconstruction. First, some new concepts are proposed, the spatial situation perception is established by 3D reconstruction of the input fusion information, 4D reconstruction completes the situation comprehension and 5D reconstruction seeks the situation prediction. The three‐level situation estimation model is optimized to a more robust flexible model. Combined with the database system, the reasoning learning mechanism, and the diversified human‐machine interface concept, a basic framework of intelligent situation awareness is constructed. Second, the basic process of intelligent situation awareness is provided. Third, based on the flexible configuration method of the situation awareness system and the concept of unmanned weight determination for the situation awareness agents, an efficiency measurement model for the intelligent situation awareness model is also constructed, and an evaluation method for the consistency of multinode intelligent situation awareness is also provided. Fourth, the paper gives a typical electromagnetic situation estimation example for a drilling platform to explain and validate the concepts. Finally, several suggestions are put forward for the next construction of an intelligent situation awareness system.

A Procedural Constructive Learning Mechanism with Deep Reinforcement Learning for Cognitive Agents

Recent advancements in AI and deep learning have created a growing demand for artificial agents capable of performing tasks within increasingly complex environments. To address the challenges associated with continuous learning constraints and knowledge capacity in this context, cognitive architectures inspired by human cognition have gained significance. This study contributes to existing research by introducing a cognitive-attentional system employing a constructive neural network-based learning approach for continuous acquisition of procedural knowledge. We replace an incremental tabular Reinforcement Learning algorithm with a constructive neural network deep reinforcement learning mechanism for continuous sensorimotor knowledge acquisition, thereby enhancing the overall learning capacity. The primary emphasis of this modification centers on optimizing memory utilization and reducing training time. Our study presents a learning strategy that amalgamates deep reinforcement learning with procedural learning, mirroring the incremental learning process observed in human sensorimotor development. This approach is embedded within the CONAIM cognitive-attentional architecture, leveraging the cognitive tools of CST. The proposed learning mechanism allows the model to dynamically create and modify elements in its procedural memory, facilitating the reuse of previously acquired functions and procedures. Additionally, it equips the model with the capability to combine learned elements to effectively adapt to complex scenarios. A constructive neural network was employed, initiating with an initial hidden layer comprising one neuron. However, it possesses the capacity to adapt its internal architecture in response to its performance in procedural and sensorimotor learning tasks, inserting new hidden layers or neurons. Experimentation conducted through simulations involving a humanoid robot demonstrates the successful resolution of tasks that were previously unsolved through incremental knowledge acquisition. Throughout the training phase, the constructive agent achieved a minimum of 40% greater rewards and executed 8% more actions when compared to other agents. In the subsequent testing phase, the constructive agent exhibited a 15% increase in the number of actions performed in contrast to its counterparts.

Make digitalized places for experimentation work: unravelling and governing transformative dynamics of FabLabs and Makerspaces

ABSTRACT While maker platforms, such as FabLabs and Makerspaces, are commonly perceived as drivers for transformative trajectories, little is known about how the dynamics of experimental collective agency are mobilizing transformative capacities that foster change on the local level. This paper contributes to the understanding of collective innovation dynamics of maker platforms that aim to support collective experimentation and translate novel practices to incumbent institutions on the local level. By drawing on the strategic niche management literature, a comparative case study of six maker platforms in Germany is conducted. Findings indicate three relevant modes of translation that could be identified: Firstly, translation by active shielding and institutionalized adaption emphasizes the role of harbouring universities in fencing off market pressures and offering events as well as workshops to regime actors. Secondly, translation by enabling learning mechanisms highlights the effect of implementing proper formats for assessment and learning on mutual translation. Thirdly, translation by sustained cooperation and coordination between maker platforms and local institutions stresses capacities for cooperation and a demand-oriented coordination between maker platforms and incumbent institutions.

Investigating boundaries and boundary crossing between mathematics and visual art teaching in a collaborative setting

Our study explores boundaries and boundary crossing between communities of mathematics teaching (MT) and visual art teaching (AT). We focus on teacher collaboration, draw on communities of practice and examine the existing boundaries and the way collaborating members handle them. We present data from 17 group meetings of secondary school art and mathematics teachers who try to develop ways of linking ΜΤ and ΑΤ. Results indicate the emerging boundaries by means of discontinuities regarding mathematical practices and tools used in MT and AT, and the disciplines’ teaching and curriculums. Via analysis of the central boundary – the analytical or visual ways of thinking – we present members’ boundary handling that indicates the development and developmental process of integrated practice. The study incorporates boundary crossing processes and learning mechanisms in which the two separate ways of thinking are gradually transformed into an integrated one and reveals possible ways of integration in teaching and learning.

Neurocognitive mechanisms of mental imagery-based disgust learning

Disgust imagery represents a potential pathological mechanism for disgust-related disorders. However, it remains controversial as to whether disgust can be conditioned with disgust-evoking mental imagery serving as the unconditioned stimulus (US). Therefore, we examined this using a conditioned learning paradigm in combination with event-related potential (ERP) analysis in 35 healthy college students. The results indicated that the initial neutral face (conditioned stimulus, CS+) became more disgust-evoking, unpleasant, and arousing after pairing with disgust-evoking imagery (disgust CS+), compared to pairing with neutral (neutral CS+) and no (CS−) imagery. Moreover, we observed that mental imagery-based disgust conditioning was resistant to extinction. While the disgust CS + evoked larger P3 and late positive potential amplitudes than CS− during acquisition, no significant differences were found between disgust CS+ and neutral CS+, indicating a dissociation between self-reported and neurophysiological responses. Future studies may additionally acquire facial EMG as an implicit index of conditioned disgust. This study provides the first neurobiological evidence that associative disgust learning can occur without aversive physical stimuli, with implications for understanding how disgust-related disorders may manifest or deteriorate without external perceptual aversive experiences, such as in obsessive-compulsive disorder (OCD).

Evolution of teachers’ and researchers’ praxeologies for designing inquiry mathematics tasks: the role of teachers’ beliefs

AbstractThis paper is focused on the collaborative work of two communities, one of teachers and one of researchers, during a teacher professional development program on the inquiry-based learning approach in mathematics, addressed to lower secondary school in-service mathematics teachers. We conceptualize the design of inquiry mathematics task as the boundary object on which the two communities work collaboratively. We aim to study the evolution of teachers’ and researchers’ meta-didactical praxeologies for designing inquiry mathematics tasks, with the Meta-Didactical Transposition framework, to understand if and how their collaboration favors their convergence toward shared components of final meta-didactical praxeologies. In the results, we show that this convergence is reached thanks to internalization processes of praxeological components for designing inquiry mathematics tasks, as a consequence of the learning mechanisms activated by the common work on the boundary object. In this paper, we address also the issue of understanding the complex relationship between teachers’ beliefs on inquiry mathematics tasks and the meta-didactical praxeologies of teachers and researchers. As a result, we propose a model in which the evolution of teachers’ beliefs is taken into account both as an agent and a consequence of the evolution of the meta-didactical praxeologies of the two communities involved in the teacher professional development program.