Article
Behavioral Sciences
Benjamin H. Paffhausen, Julian Petrasch, Uwe Greggers, Aron Duer, Zhengwei Wang, Simon Menzel, Peter Stieber, Karen Haink, Morgan Geldenhuys, Jana Cavojska, Timo A. Stein, Sophia Wutke, Anja Voigt, Josephine Coburn, Randolf Menzel
Summary: Honeybees serve as indicators of ecosystem health through social signals and the emission of characteristic electrostatic fields. By analyzing these signals along with physical measurements, the overall condition of the colony can be quantified, providing valuable insights into the health of the entire ecosystem.
FRONTIERS IN BEHAVIORAL NEUROSCIENCE
(2021)
Article
Telecommunications
Huan Luo, Kaitian Cao, Yongpeng Wu, Xiaoming Xu, Yuan Zhou
Summary: In this paper, a new hybrid priority queuing model is proposed and a deep Q-network-based algorithm is designed to optimize the latency performance of spectrum handoffs in cognitive radio networks. The transfer learning method is introduced to accelerate the learning process, and simulation results show that the proposed method outperforms conventional reinforcement learning-based approaches in terms of latency performance.
IEEE COMMUNICATIONS LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Laurens Bliek, Arthur Guijt, Rickard Karlsson, Sicco Verwer, Mathijs de Weerdt
Summary: The article highlights the lack of standardization in benchmarking surrogate algorithms on real-life, expensive, black-box objective functions. It introduces a new benchmark library called EXPObench, which provides a comprehensive comparison of six different surrogate algorithms on four expensive optimization problems from different real-life applications. The study reveals new insights on the importance of exploration, objective evaluation time, and the choice of model, and offers practical recommendations for selecting surrogate algorithms in different situations.
APPLIED SOFT COMPUTING
(2023)
Article
Engineering, Industrial
Bing Chen, Ruibin Bai, Jiawei Li, Yueni Liu, Ning Xue, Jianfeng Ren
Summary: This paper discusses the current situation and limitations of current methods for addressing uncertainties in real-life optimization problems. The authors propose a novel framework that combines mathematical models and machine learning modules to overcome these limitations and demonstrate its practicality and feasibility through real-life and artificial bus scheduling instances. The proposed framework represents the first multi-objective bus-headway-optimisation method for non-timetabled bus schedules with major practical constraints being considered.
INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH
(2023)
Article
Computer Science, Interdisciplinary Applications
Bo Li, Fen Hou, Hongwei Ding, Hao Wu
Summary: This paper proposes a general architecture for building crowdsensing-based parking guiding system and investigates its performance in various parking environments. Experimental results show that the crowdsensing-based parking prediction method can lead to 30.91% or more relative improvement on average estimation error than steady-state prediction.
COMPUTERS & INDUSTRIAL ENGINEERING
(2021)
Article
Computer Science, Artificial Intelligence
Pongchanun Luangpaiboon, Pasura Aungkulanon, Roberto Montemanni
Summary: This paper presents a centralised method based on evolutionary computation elements and a novel elevator kinematic optimisation algorithm to solve the manufacturing parameter design problem. The method, called compromise multi response surface optimisation method, simultaneously produces multiple flexi-design plans to reduce the number of experiments and is used in the design phase of Taguchi arrays. The method also allows for choosing parameter levels during the analysis phase without new individual fitting sets using available information. Benchmarking results from industrial application indicate that the proposed method can lead to better optimal solutions for complex processes such as resistance welding compared to the conventional Taguchi method.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Automation & Control Systems
Wankun Sirichotiyakul, Aykut C. Satici
Summary: This paper introduces a systematic approach to design controllers for underactuated mechanical systems based on interconnection and damping assignment. By exploiting the universal approximation capability of neural networks, the solutions to the required PDEs are automatically discovered without destroying the passivity structure of the system.
INTERNATIONAL JOURNAL OF CONTROL
(2023)
Article
Computer Science, Artificial Intelligence
Maged Mamdouh, Mostafa Ezzat, Hesham Hefny
Summary: Accurate aircraft delay prediction is of significant importance in the aviation sector to reduce losses and improve passenger loyalty. This paper proposes an integrated network framework called 'Attention-based Bidirectional long short-term memory' (ATT-BI-LSTM) for flight delay prediction using machine learning techniques. The simulation results show that the proposed framework outperforms other models, indicating its potential in real-time monitoring of flight delays.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Green & Sustainable Science & Technology
Ziming Wang, Chaohao Liao, Xu Hang, Lishuai Li, Daniel Delahaye, Mark Hansen
Summary: Predicting flight delays has been a major research topic, with most focus on short-term prediction. This paper proposes machine learning methods to predict the distribution of flight delays and validates them using empirical data from Guangzhou Baiyun International Airport. The results show that the proposed methods can accurately predict departure and arrival delays at the strategic stage. This research provides an important tool for airports and airlines.
Article
Computer Science, Artificial Intelligence
Xiaodong Feng, Zhen Liu, Wenbing Wu, Wenbo Zuo
Summary: The rapid development of social recommendation in recent years has greatly improved the performance of recommender systems, especially for the cold start problem. However, existing techniques based on matrix factorization do not effectively capture the complex nonlinear relationships between users and items, as well as between users themselves. To address this, deep learning is employed to model the social network-enhanced collaborative filtering problem. By simultaneously modeling the social and item domain interactions, the proposed SoNeuMF framework shows significant improvements in recommendation accuracy compared to state-of-the-art methods, as demonstrated by comprehensive experiments on real-world datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Nanoscience & Nanotechnology
Matthew Guziewski, David Montes de Oca Zapiain, Remi Dingreville, Shawn P. Coleman
Summary: This study characterizes the energy and strength of silicon carbide grain boundaries using a combination of high-throughput atomistic simulations, macroscopic and microscopic descriptors, and machine-learning techniques. Results show that while microscopic descriptors are more effective in describing grain boundary energy, a combination of macroscopic and microscopic descriptors can accurately predict grain boundary strength.
ACS APPLIED MATERIALS & INTERFACES
(2021)
Article
Psychology, Multidisciplinary
Miao Li, Ying Hua
Summary: This study investigates the impact of the social presence generated by real-time interactions on consumers' purchase intention in the online shopping environment. The results show that social presence positively affects consumers' exploratory and exploitative learning, leading to cognitive and affective appraisal, which ultimately influences purchase intention. The findings provide valuable insights for brand managers and retailers in live streaming marketing.
FRONTIERS IN PSYCHOLOGY
(2022)
Article
Mathematics
Emanuel Vega, Ricardo Soto, Broderick Crawford, Javier Pena, Carlos Castro
Summary: The study introduces a novel optimisation framework called LB2, focusing on predicting better movements for improved performance. Testing with movement operators of a spotted hyena optimiser, the hybrid approach is found to be competitive compared to state-of-the-art algorithms and sequential parameter optimisation methods in solving benchmark functions.
Article
Computer Science, Artificial Intelligence
Jun Yan, Qi Zhang, Qi Xu, Zhirui Fan, Haijiang Li, Wei Sun, Guangyuan Wang
Summary: The paper introduces a deep learning approximate algorithm based on convolutional neural networks for real-time topology optimization, which can produce high-precision topological prediction results with a relatively small number of samples and has higher real-time performance compared to traditional topology optimization methods.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Engineering, Environmental
Sina Moradi, Amr Omar, Zhuoyu Zhou, Anthony Agostino, Ziba Gandomkar, Heriberto Bustamante, Kaye Power, Rita Henderson, Greg Leslie
Summary: Four machine learning algorithms were used to predict the performance of a multi-media filter. Random Forest with a 1-day time lag provided the highest reliability in predicting unit filter run volume. It can help water treatment operators make real-time decisions by warning of potential turbidity breakthrough.