Recurrent neural network-based sequence-to-sequence (S2S) models have been extensively applied for multistep-ahead time-series forecasting. These models typically involve a decoder trained using either its previous forecasts or the actual observed values as the decoder inputs. However, relying on self-generated predictions can lead to the rapid accumulation of errors over multiple steps, while using the actual observations introduces exposure bias as these values are unavailable during the inference stage. In this regard, this study proposes a novel training approach called reinforced decoder, which introduces auxiliary models to generate alternative decoder inputs that remain accessible when extrapolating. In addition, a reinforcement learning algorithm is utilized to dynamically select the optimal inputs to improve accuracy. Comprehensive experiments demonstrate that our approach outperforms representative training methods over several datasets. Compared to the best baseline, the proposed method based on long short-term memory yields average relative improvements of 5.98% in root mean square error, 4.39% in mean absolute percentage error, and 7.63% in$\text{R}^{2}$. Furthermore, the proposed approach can also achieve promising performance when generalized to Transformer-based S2S forecasting models.

Group decision-making (GDM) is a crucial approach to ensuring the scientific nature and impartiality of decisions. However, strategic manipulative behaviors driven by self-interested motives often undermine the fairness and effectiveness of decision outcomes, leading to results that deviate from expectations. While most prior studies have focused on theoretical analysis, there remains a significant gap in effective measures to prevent such manipulative behaviors. Moreover, current consensus models predominantly emphasize cost optimization, with less attention paid to the acceptability of feedback. To address these challenges, this study introduces an optimal consensus adjustment mechanism based on the Nash bargaining (NB) solution, aiming to prevent manipulation and self-interested behaviors in GDM. Specifically, we first analyze the opinion manipulation problem within the framework of the minimum adjustment consensus model (MACM). We then construct the Nash product to mitigate the risk of weight manipulation. Subsequently, we examine the nonuniqueness issue in the allocation of minimal total consensus adjustments from the perspective of cooperative game theory. Building on this, we incorporate regret theory to characterize the risk aversion and loss sensitivity of decision-makers (DMs) and propose a consensus adjustment mechanism based on the NB game. Finally, we establish three novel optimization methods to allocate optimal individual consensus adjustments. Case studies and comparative experiments demonstrate the superiority of these methods.

While the elderly population is growing rapidly, acceptance and use of m-government services by them are far below expectation. Previous studies on acceptance and use of m-government services have predominantly focused on younger citizens with skills and experience of information technologies. Drawing upon the dual factor model, this study investigates the enablers and inhibitors of the elderly's m-government service adoption behavior. Four constructs from the unified theory of acceptance and use of technology (UTAUT), namely, performance expectancy, effort expectancy, facilitating conditions, social influence; and self-actualization are treated as enablers, while user resistance to change, technology anxiety, and declining physiological conditions are regarded as inhibitors. Results show that adoption of m-government by the elderly is significantly influenced by all tested enablers and inhibitors, except for social influence. This study contributes by providing an integrative model of technology acceptance for the elderly along with practical implications for policy makers.

Quantization is known as an effective compression method for deploying neural networks on mobile devices. However, most existing works train from scratch a quantized network with universal bitwidth for all layers, making it hard to find the optimal trade-off between compression ratio and inference accuracy. In this paper, we propose a novel post-training quantization approach which derives a flexible bitwidth scheme. Our algorithm progressively downgrades bitwidth of chosen layer in the network and performs feature map alignment with pre-trained model. The algorithm comprises a meter of layer sensitivity and an iterative quantizer. Specifically, the meter dynamically estimates for every layer the error of quantization on its output feature map, meanwhile the error serves as an objective function to be minimized by the quantizer. Extensive experiments on CIFAR-10 and ImageNet ILSVRC2012 datasets demonstrate that the proposed approach achieves impressive results for mainstream neural networks.

Twenty-First Century Education is a design of instructional culture that empower learner-centered through the philosophy of ‘Less teaching but more learning’. Due to the development of technology enhance learning in developing countries such as Thailand, online learning is rapidly growing in the electronic learning market. ClassStart is a learning management system developed to support Thailand's educational management and to promote the student-centred learning processes. The study of online learning acceptance is primarily required to successfully achieve online learning system development. However, the behavioural intention of students to use online learning systems has not been well examined, in particular, by focusing specific but representative applications such as ClassStart in this study. This research takes the usage of ClassStart as research scenario and investigates the individual acceptance of technology through the Unified Theory of Acceptance and Use of Technology, as well as technological quality through the Delone and McLean IS success model. The Partial Least Squares method, a statistics analysis technique based on the Structural Equation Model (SEM), was used to analyze the data. It was found that performance expectancy, social influence, information quality and system quality have the significant effect on intention to use ClassStart.

Malicious web domains represent a big threat to web users' privacy and security. With so much freely available data on the Internet about web domains' popularity and performance, this study investigated the performance of well-known machine learning techniques used in conjunction with this type of online data to identify malicious web domains. Two datasets consisting of malware and phishing domains were collected to build and evaluate the machine learning classifiers. Five single classifiers and four ensemble classifiers were applied to distinguish malicious domains from benign ones. In addition, a binary particle swarm optimisation (BPSO) based feature selection method was used to improve the performance of single classifiers. Experimental results show that, based on the web domains' popularity and performance data features, the examined machine learning techniques can accurately identify malicious domains in different ways. Furthermore, the BPSO-based feature selection procedure is shown to be an effective way to improve the performance of classifiers.

Opposition-based Learning (OBL) has been reported with an increased performance in enhancing various optimization approaches. Instead of investigating the opposite point of a candidate in OBL, this study proposed a partial opposition-based learning (POBL) schema that focuses a set of partial opposite points (or partial opposite population) of an estimate. Furthermore, a POBL-based adaptive differential evolution algorithm (POBL-ADE) is proposed to improve the effectiveness of ADE. The proposed algorithm is evaluated on the CEC2014's test suite in the special session and competition for real parameter single objective optimization in IEEE CEC 2014. Simulation results over the benchmark functions demonstrate the effectiveness and improvement of the POBL-ADE compared with ADE.

Multistep-ahead time series prediction is one of the most challenging research topics in the field of time series modeling and prediction, and is continually under research. Recently, the multiple-input several multiple-outputs (MISMO) modeling strategy has been proposed as a promising alternative for multistep-ahead time series prediction, exhibiting advantages compared with the two currently dominating strategies, the iterated and the direct strategies. Built on the established MISMO strategy, this paper proposes a particle swarm optimization (PSO)-based MISMO modeling strategy, which is capable of determining the number of sub-models in a self-adaptive mode, with varying prediction horizons. Rather than deriving crisp divides with equal-size s prediction horizons from the established MISMO, the proposed PSO-MISMO strategy, implemented with neural networks, employs a heuristic to create flexible divides with varying sizes of prediction horizons and to generate corresponding sub-models, providing considerable flexibility in model construction, which has been validated with simulated and real datasets.

Recommendation systems are usually used in E-commerce sites to suggest products to their customers and to provide consumers with information to help them decide which products to be purchased. Many different approaches including web usage mining have been applied to the basic problem of developing accurate and efficient recommendation systems. This paper presents the application of Profile Aggregations based on Clustering Transactions (PACT), a widely used techniques in web usage mining, in designing a BtoC E-commerce recommendation system.