Les faisans ne sont pas tous égaux face au renard, les plus doués en termes de mémoire ayant le plus grand territoire et surtout les plus grandes chances de survie, selon une étude lundi.

En théorie, rien de nouveau, la taille du territoire de la plupart des animaux serait liée à ses capacités cognitives, ne serait-ce que pour se souvenir de ses limites. Mais cela reste difficile à prouver "parce qu'ils peuvent avoir d'autres raisons de se limiter à un petit territoire", par exemple s'ils y trouvent suffisamment de ressources, explique le biologiste de l'évolution Robert Heathcote, de l'Université de Bristol.

Pour en avoir le cœur net, une équipe de l'Université britannique d'Exeter, et d'universités néerlandaise et israéliennes, a mené une expérience grandeur nature dans une forêt du Devon, dans le sud-ouest de l'Angleterre.

Avant d'y être lâchés, 126 faisans élevés en captivité ont subi sur quelques semaines trois tests jaugeant leur capacités cognitives, et notamment deux types de mémoire spatiale.

La mémoire dite de travail, qui est de court terme, permet à un individu de se souvenir que s'il a trouvé un ver de terre à un endroit, il ne sert à rien d'y retourner cinq minutes plus tard. La deuxième, dite mémoire de référence spatiale, de plus long terme, permet au faisan de se souvenir d'un trajet même après plusieurs jours.

L'étude publiée dans Nature Ecology & Evolution établit que c'est cette mémoire de référence spatiale qui dicte la taille du territoire d'un faisan. Ce territoire, "qui est la zone où il passe l'essentiel de son temps, est aussi celui qu'il connaît le mieux", selon M. Heathcote. Son étendue va de moins de cent mètres de long et jusqu'à un kilomètre carré.

- Territoires de la mort -

En l'espace de six mois, les chercheurs ont enregistré la prédation de 45 faisans, tous sous les crocs de renards roux. Chaque volatile était équipé d'une minuscule balise d'une dizaine de grammes, conçue par les chercheurs israéliens, permettant sa localisation quasiment en temps réel.

"Ce qui a permis de savoir quand la trajectoire de la balise n'était plus celle du faisan, mais était devenue celui du renard", explique en souriant M. Heathcote à l'AFP. La déambulation prudente du volatile se muait, une fois saisi dans la mâchoire de son prédateur, en trajectoire rectiligne, rapide et lointaine du renard, vers un endroit où dévorer tranquillement sa proie.

Les faisans les plus susceptibles de terminer leur existence de cette manière étaient ceux ayant une piètre mémoire de référence spatiale. Leur fin était aussi beaucoup plus probable aux frontières de leur territoire. "La connaissance d'une zone aide le faisan à rester vivant", et inversement, selon le Dr Joah Madden, de l'Université d'Exeter, cité dans un communiqué.

Même dans les zones de chasse préférées des renards, que M. Heathcote a baptisé "territoires de la mort", les chances de survie d'un faisan dépendent avant tout de son expérience du terrain. Les plus habiles n'évitent pas la zone de la mort, mais "avec le temps ils peuvent apprendre quelles sont les voies les plus rapides et les plus sûres pour échapper à une attaque".

Pour les faisans qui échappent aux crocs, reste le risque de finir criblé de plomb par l'homme.

(Belga) Un Américain, qui avait été immortalisé les pieds sur une table dans le bureau de la cheffe démocrate Nancy Pelosi lors de l'assaut sur le Capitole, a été reconnu coupable lundi de plusieurs délits.

Après une courte délibération, les jurés ont déclaré Richard Barnett, 62 ans, coupable, entre autres, d'entrave à une procédure officielle, vol et intrusion dans un bâtiment officiel avec une arme dangereuse (un bâton de marche capable d'envoyer des décharges électriques). Le 6 janvier 2021, il avait envahi, comme des centaines de partisans de l'ex-président républicain Donald Trump, le siège du Congrès au moment où les élus certifiaient la victoire du démocrate Joe Biden à la présidentielle. Il avait été photographié par l'AFP dans le bureau de la cheffe de la chambre des représentants, Nancy Pelosi, les pieds sur un meuble. Le cliché avait fait le tour du monde et permis à la police de l'interpeller rapidement. Selon le dossier d'accusation, ce partisan de la mouvance complotiste Qanon avait laissé un message insultant à la démocrate et volé une enveloppe qu'elle avait signée. Pendant son procès, il s'était montré défiant, assurant avoir été "poussé à l'intérieur" du Capitole par la foule. Sa peine sera prononcée en mai. En attendant, il reste assigné à résidence avec un bracelet électronique. En deux ans d'enquête, plus de 950 participants à cette attaque ont été arrêtés, et près de 200 condamnés à des peines de prison. (Belga)

(Belga) Le conseil communal de la Ville de Charleroi a adopté lundi une motion faisant de Charleroi "une ville antifasciste" et consacrant l'existence d'une "coalition antifasciste" composée des partis politiques carolos, des syndicats, d'associations et de membres de la société civile.

Cette "coalition antifasciste" est le fruit de discussions entamées dans un contexte de montée générale des idées d'extrême droite et à la suite des incidents qui sont survenus le 25 janvier 2020 à Charleroi à l'occasion d'une mobilisation d'un front antifasciste contre la tenue dans la métropole d'une réunion d'un nouveau parti d'extrême droite. Ce jour-là, selon les manifestants antifascistes, la police avait fait usage contre eux de sprays, d'autopompes et de coups de matraques même pour les disperser. Ce qui avait provoqué un certain émoi, y compris au sein de la classe politique carolo. La motion donne à la coalition antifasciste quelques objectifs généraux, comme celui "d'empêcher par tous les moyens légaux la diffusion de propos incitant à la haine, au racisme, à l'antisémitisme, au sexisme, à la discrimination relative à l'orientation sexuelle, ouvertement fasciste et xénophobe, sur le territoire de Charleroi" ou celui de relayer l'information "lorsqu'elle concerne un événement susceptible d'inciter à la haine, au racisme, à l'antisémitisme, au sexisme, ouvertement fasciste et xénophobe". (Belga)

(Belga) Quatre membres de la milice d'extrême droite "Oath Keepers" ont été reconnus coupables lundi de sédition pour leur rôle dans l'assaut du Capitole, à l'issue du second procès organisé sur ce chef d'accusation extrêmement rare.

Depuis l'attaque du 6 janvier 2021, plus de 950 partisans de l'ex-président républicain Donald Trump ont été arrêtés et inculpés pour avoir semé le chaos dans le siège de la démocratie américaine. Parmi eux, seuls 14 militants de groupuscules d'extrême droite - neuf membres des "Oath Keepers" et cinq "Proud Boys" - ont été accusés de "sédition", un chef passible de 20 ans de prison qui implique d'avoir planifié l'usage de la force pour s'opposer au gouvernement. Faute de place suffisante dans le tribunal fédéral de Washington, la justice a organisé le procès des Oath Keepers, accusés de s'être entraînés et armés pour l'occasion, en deux temps. Un premier procès s'est conclu fin novembre par un verdict mitigé: le fondateur de cette milice, Stewart Rhodes, et un responsable local ont été déclarés coupables de sédition, mais leurs trois co-accusés ont été acquittés sur ce chef. Lundi, à l'issue du second procès, les jurés ont jugé coupables les quatre derniers Oath Keepers, des hommes âgés de 38 à 64 ans décrits comme de dangereux "traîtres" par l'accusation, mais comme des "fanfarons" par leurs avocats. Le procès des Proud Boys, dont leur leader Enrique Tarrio, s'est ouvert en décembre et était toujours en cours lundi, dans le même tribunal. (Belga)

(Belga) Sept personnes ont été tuées lundi lors d'une double fusillade près de San Francisco, en Californie, ont indiqué les médias américains sur la base des déclarations de la police locale.

Le suspect a été arrêté, a annoncé sur Twitter le bureau du shérif du comté de San Mateo, qui comprend la ville de Half Moon Bay où ont eu lieu les drames. "Il n'y a plus de danger pour la population à cette heure", a-t-il assuré. Les deux fusillades sont intervenues dans des exploitations agricoles proches l'une de l'autre, ont précisé les médias. Cette nouvelle tuerie intervient moins de 48 heures après qu'un tireur a tué 11 personnes dans un club de danse près de Los Angeles. (Belga)

L'information est ￜ lire sur le site de Mediapart : le 15 mai, alors que l'Assemblￜe est en train de voter la poursuite de l'ￜtat d'urgence sanitaire, les dￜputￜs de la Rￜpublique en...

C'est surrￜaliste. Les services de Matignon ont dￜcouvert fin mars que des stocks de masques pￜrimￜs continuaient ￜ ￜtre brￜlￜs pendant l'ￜpidￜmie alors que certains...

Vous n'avez pas eu le temps de lire Le Canard enchaￜnￜ, Mediapart, Le Monde, Arrￜt sur images et tous les autres titres de presse ? On s'en charge pour vous.

Hasard du calendrier ou volontￜ de prￜserver le nouveau Premier ministre ? Selon Mediapart, une enquￜte judiciaire ouverte par le parquet de Perpignan, potentiellement gￜnante pour Jean Castex, a ￜtￜ...

Des files d'attente, la nuit, devant les prￜfectures, pour tenter d'obtenir un rendez-vous afin de demander ou de renouveler un titre de sￜjour. C'ￜtait la rￜalitￜ au dￜbut des annￜes...

L'information est passï¿œe inaperï¿œue mais elle ne manque pas de sel. Alors que la gestion de la crise du covid-19 par le gouvernement est trï¿œs contestï¿œe (au retard dans la livraison de masques s'ajoute...

La crise ? Quelle crise ? Alors que le gouvernement prￜpare l'opinion ￜ une longue pￜriode d'inflation et de hausse des prix de l'ￜnergie en raison de la guerre en Ukraine, tout ne va pas si mal sur le plan...

Through the unparalleled espousal of theory of planned behaviour, this study intends to significantly add to the current knowledge on social networking sites (SNS) in <i>eWOM</i> information and its role in defining intentions to buy green products. In specie, this study seeks to first investigate the part played by <i>attitude towards SNS information</i> in influencing the <i>acceptance of SNS information</i> and then by <i>acceptance of SNS information</i> in effecting the <i>green purchase intention</i>. Besides this, it also aims to analyse the influence exerted by first <i>credibility of SNS information</i> on <i>acceptance of SNS information</i> and then by <i>acceptance of SNS information</i> on <i>green purchase intention</i>. In doing so, it also examines how well the age of the SNS users moderates all these four associations.

Aiming at the problems of high error rate, low evaluation accuracy and low investment return in traditional methods, a random forest-based e-bank investment risk evaluation method is proposed. First, establish a scientific e-bank investment risk evaluation index system. Then, G1-COWA combined weighting method is used to calculate the weights of each index. Finally, the e-bank investment risk evaluation index data is taken as the input vector, and the e-bank investment risk evaluation result is taken as the output vector. The random forest model is established and the result of e-banking investment risk evaluation is obtained. The experimental results show that the maximum relative error rate of this method is 4.32%, the evaluation accuracy range is 94.5~98.1%, and the maximum return rate of e-banking investment is 8.32%. It shows that this method can accurately evaluate the investment risk of electronic banking.

The operation of supply chain finance faces various risks, therefore, studying the operational risks of supply chain finance and corresponding preventive measures is of great significance. Firstly, classify the types of operational risks in supply chain finance. Secondly, based on the risk classification results, the decision tree method is used to evaluate the operational risks of supply chain finance. Finally, based on the risk assessment results, targeted risk prevention measures for supply chain finance operations are proposed, such as strengthening supplier management, optimising logistics and warehouse management, risk analysis and monitoring, and strengthening information security and data protection. The case analysis results show that the accuracy of the evaluation results of this method is higher, and the risk coefficient has been significantly reduced after applying this method, indicating that it can effectively reduce supply chain risk.

The current advertising push methods have low accuracy and poor advertising conversion effects. Therefore, a Weibo marketing advertising push method based on social network data mining is studied. Firstly, establish a social network graph and use graph clustering algorithm to mine the association relationships of users in the network. Secondly, through sparsisation processing, the association between nodes in the social network graph is excavated. Then, evaluate the tightness between user preferences and other nodes in the social network, and use the TF-IDF algorithm to extract user interest features. Finally, an attention mechanism is introduced to improve the deep learning model, which matches user interests with advertising domain features and outputs push results. The experimental results show that the push accuracy of this method is higher than 95%, with a maximum advertising click through rate of 82.7% and a maximum advertising conversion rate of 60.7%.

In order to solve the problems of long prediction consumption time and many prediction iterations existing in traditional prediction models, an e-commerce growth prediction model based on grey Markov chain is proposed. The Scrapy crawler framework is used to collect a variety of e-commerce data from e-commerce websites, and the feedforward neural network model is used to clean the collected data. With the cleaned e-commerce data as the input vector and the e-commerce growth prediction results as the output vector, an e-commerce growth prediction model based on the grey Markov chain is built. The prediction model is improved by using the background value optimisation method. After training the model through the improved particle swarm optimisation algorithm, accurate e-commerce growth prediction results are obtained. The experimental results show that the maximum time consumption of e-commerce growth prediction of this model is only 0.032, and the number of iterations is small.

To reduce the location cost of cross-border e-commerce logistics sites, this article proposes a multi-logistics site location method based on return uncertainty. Firstly, a site selection model is established with the objective function of minimising site construction costs, transportation costs, return costs, and operating costs, and the constraint conditions of return recovery costs and delayed pick-up time; Then, using the Monte Carlo method to simulate the number of returned items, and using an improved chicken swarm algorithm based on simulated annealing, the cross-border e-commerce multi-logistics site location model is solved to complete the location selection. Experimental results show that this method can effectively reduce the related costs of cross-border e-commerce multi-logistics site selection. After applying this method, the total cost of multi-logistics site selection is 19.4 million yuan, while the total cost of the five comparative methods exceeds 20 million yuan.

With the rapid development and application of driverless technology, the number and location of vehicles, the channel and bandwidth of wireless network are time-varying, which leads to the increase of offloading delay and energy consumption of existing algorithms. To solve this problem, the vehicle terminal task offloading decision problem is modelled as a Markov decision process, and a task offloading algorithm based on DDQN is proposed. In order to guide agents to quickly select optimal strategies, this paper proposes an offloading mechanism based on task feature. In order to solve the problem that the processing delay of some edge server tasks exceeds the upper limit of their delay, a task scheduling mechanism based on buffer delay is proposed. Simulation results show that the proposed method has greater performance advantages in reducing delay and energy consumption compared with existing algorithms.

To achieve the benefits as much as possible, it is required to identify the available PEV capacity and prepare scheduling plans based on that. The analysis revealed that the risk-based scheduling of the microgrid could reduce the financial risk completely from $9.89 to $0.00 and increases the expected operation cost by 24% from $91.38 to $112.94, in turn. This implies that the risk-averse decision-maker tends to spend more money to reduce the expected risk-in-cost by using the proposed downside risk management technique. At the end, by the help of fuzzy satisfying method, the suitable risk-averse strategy is determined for the studied case.

Image forgery detection techniques refer to the process of detecting manipulated or altered images, which can be used for various purposes, including malicious intent or misinformation. Image forgery detection is a crucial task in digital image forensics, where researchers have developed various techniques to detect image forgery. These techniques can be broadly categorised into active, passive, machine learning-based and hybrid. Active approaches involve embedding digital watermarks or signatures into the image during the creation process, which can later be used to detect any tampering. On the other hand, passive approaches rely on analysing the statistical properties of the image to detect any inconsistencies or irregularities that may indicate forgery. In this paper for the detection of scaling and cropping attack a deep learning method has been proposed using ResNet. The proposed method (Res-Net-Adam-Adam) is able to achieve highest amount of accuracy of 99.14% (0.9914) while detecting fake and real images.

Screen-shooting will lead to information leakage. Anti-screen-shooting watermark, which can track the leaking sources and protect the copyrights of images, plays an important role in image information security. Due to the randomness of shooting distance and angle, more robust watermark algorithms are needed to resist the mixed attack generated by screen-shooting. A robust digital watermarking algorithm that is resistant to screen-shooting is proposed in this paper. We use improved Harris-Laplace algorithm to detect the image feature points and embed the watermark into the feature domain. In this paper, all test images are selected on the dataset USC-SIPI and six related common algorithms are used for performance comparison. The experimental results show that within a certain range of shooting distance and angle, this algorithm presented can not only extract the watermark effectively but also ensure the most basic invisibility of watermark. Therefore, the algorithm has good robustness for anti-screen-shooting.

The current study has outlined slow fashion (SF) research trends and created a future research agenda for this field. It is a thorough analysis of the literature on slow fashion. Numerous bibliometric features of slow fashion have been discussed in the paper. This study comprises 182 research articles from the Scopus database. The database was utilised for bibliometric analysis. To identify certain trends in the area of slow fashion, a bibliometric study is done. For bibliometric analysis, the study employed R-software (the Biblioshiny package). Here, VOSviewer software is used to determine the co-occurrence of authors, countries, sources, etc. The study has outlined the gap that still exists in the field of slow fashion. Here, the research outcome strengthens the domain of slow fashion for sustainable consumption. The study findings will be useful for policymakers, industry professionals, and researchers.

The video game industry (VGI) has evolved considerably, transitioning from a niche market to a substantial sector. The VGI's magnitude and the societal implications tied to video game consumption have naturally piqued the interest of scholars in marketing and consumer behaviour. This research serves a dual purpose: firstly, it consolidates existing VG literature by evaluating articles, concepts, and methodologies, systematically tracing their evolution; secondly, it outlines potential directions and implications for forthcoming research. Within this literature, a predominant focus lies on articles investigating purchase decisions concerning VGs, followed by those exploring the integration of video game consumption into broader social contexts. Notably, a limited number of articles delve into player-game interactions and experiences within gaming worlds. This imbalance can be attributed to the fact that such inquiries are often suited to psychology and multidisciplinary journals, while the marketing discipline has predominantly addressed the VGI from a marketing management standpoint.

This study investigates the effects of social media marketing efforts (SMME) on smartphone brand identification, attachment, association, and willingness to pay premium prices. A survey of 320 smartphone users who followed official social media handles managed by smartphone companies was conducted for this purpose. PLS-SEM was used to analyse the collected data. The findings demonstrated importance of SMME dimensions. According to the study's findings, the smartphone brand's SMMEs had significant impact on brand identification, brand association, and brand attachment. The results revealed that SMMEs had significant impact on willingness to pay the premium price. The findings also show that brand identification, attachment, and association mediated the relationship between SMMEs and willingness to pay a premium price. The findings of this study will be useful in developing social media marketing strategies for smartphones. This study demonstrates the use of social media marketing to promote mobile phones, particularly in emerging markets.

Despite the considerable growth potential predicted for mobile games, little research explored what motivates users to be sticky and make purchases in the mobile game context. Drawing on uses and gratifications theory (UGT), this study evaluates the influencing effects of players' characteristics (i.e., individual gratification and individual situation) and the mobile game structure (i.e., presence and governance) on players' mobile game behaviour (i.e., stickiness and purchase intention). Specifically, the model was extended with factors of the individual situation and governance. After surveying 439 samples, the research model was examined using the Partial least squares structural equation modelling (PLS-SEM) approach. The results indicate that stickiness is a crucial antecedent for users' in-game purchase intention. The individual situation plays an essential role in influencing user gratification, and individual gratification is the most vital criterion affecting stickiness. Finally, except for incentives, presence, and integration positively affect stickiness. This study provides further insights into both mobile game design and governance strategies.

This paper presents a Scalable Video Coding (SVC) system over multicast Software-Defined Networking (SDN), which focuses on, transmission management for the sender-receiver model. Our approach reduces bandwidth usage by allowing the receiver to select various video resolutions in a multicast group, which helps avoid a video freezing issue during bandwidth congestion. Moreover, the SVC Multiple Sessions Transmission Bandwidth thresholds Quantised Level Balance (SVC-MST BWQLB) routes different layers of the SVC stream using distinct paths and reduces storage space and bandwidth congestion problems in different video resolutions. The experimental results show that the proposed model provides better display quality than the traditional Open Shortest Path First (OSPF) routing technique. Furthermore, it reduced transmission delays by up to 66.64% for grouped resolutions compared to SVC-Single Session Transmission (SVC-SST). Additionally, the modified Real-time Transport Protocol (RTP) header and the sorting buffer for SVC-MST are proposed to deal with the defragmentation problem.

Differential privacy is usually applied to location privacy protection scenarios, which confuses real data by adding interference noise to location points to achieve the purpose of protecting privacy. However, this method can result in a significant amount of redundant noisy data and impact the accuracy of the location. Considering the security and practicability of location data, an effective differential privacy protection method of location data based on perturbation loss constraint is proposed. After applying the Laplace mechanism under the condition of differential privacy to perturb the location data, the Savitzky-Golay filtering technology is used to correct the data with noise, and the data with large deviation and low availability is optimised. The introduction of Savitzky-Golay filtering mechanism in differential privacy can reduce the error caused by noise data while protecting user privacy. The experiments results indicate that the scheme improves the practicability of location data and is feasible.

In order to further improve the teaching quality of multimedia teaching in school daily teaching, a classroom facial expression emotion recognition model is proposed based on convolutional neural network. VGGNet and CliqueNet are used as the basic expression emotion recognition methods, and the two recognition models are fused while the attention module CBAM is added. Simulation results show that the designed classroom face expression emotion recognition model based on V-CNet has high recognition accuracy, and the recognition accuracy on the test set reaches 93.11%, which can be applied to actual teaching scenarios and improve the quality of classroom teaching.

Aiming at the problem of aggravation of traffic congestion caused by unstable signal control of traffic signal control system, the Multi-Agent Deep Deterministic Policy Gradient-based Traffic Cyclic Signal (MADDPG-TCS) control algorithm is used to control the time and data dimensions of the signal control scheme. The results show that the maximum vehicle delay time and vehicle queue length of the proposed algorithm are 11.33 s and 27.18 m, which are lower than those of the traditional control methods. Therefore, this method can effectively reduce the delay of traffic signal control and improve the stability of signal control.

To improve the effectiveness of online training for music education, it was investigated how to improve the pulse-coupled neural network in image processing for spectral image segmentation. The study proposes a two-scale descent method to achieve oblique spectral correction. Subsequently, a convolutional neural network was optimised using a two-channel feature fusion recognition network for music theory notation recognition. The results showed that this image segmentation method had the highest accuracy, close to 98%, and the accuracy of spectral tilt correction was also as high as 98.4%, which provided good image pre-processing results. When combined with the improved convolutional neural network, the average accuracy of music theory symbol recognition was about 97% and the highest score of music majors was improved by 16 points. This shows that the method can effectively improve the teaching effect of online training in music education and has certain practical value.

Physical Layer Security (PLS) and relay technology have emerged as viable methods for enhancing the security of wireless networks. Relay technology adoption enhances the extent of coverage and enhances dependability. Moreover, it can improve the PLS. Choosing relay and jammer nodes from the group of intermediate nodes effectively mitigates the presence of powerful eavesdroppers. Current methods for Joint Relay and Jammer Selection (JRJS) address the optimisation problem of achieving near-optimal secrecy. However, most of these techniques are not scalable for large networks due to their computational cost. Secrecy will decrease if eavesdroppers are aware of the relay and jammer intermediary nodes because beamforming can be used to counter the jammer. Consequently, this study introduces a multi-agent Q-learning-based PLS-enhanced secured joint relay and jammer in dual-hop wireless cooperative networks, considering the existence of several eavesdroppers. The performance of the suggested algorithm is evaluated in comparison to the current algorithms for secure node selection. The simulation results verified the superiority of the proposed algorithm.

It is inevitable for basketball players to have physical injury in sports. Reducing basketball injury is one of the main aims of the study of basketball. In view of this, this paper proposes a monocular vision and FMS injury prediction model for basketball players. Aiming at the limitations of traditional FMS testing methods, this study introduces intelligent machine learning methods. In this study, random forest algorithm was introduced into OpenPose network to improve model node occlusion, missed detection or false detection. In addition, to reduce the computational load of the network, the original OpenPose network was replaced by a lightweight OpenPose network. The experimental results show that the average processing time of the proposed model is about 90 ms, and the output video frame rate is 10 frames per second, which can meet the real-time requirements. This study analysed the students participating in the basketball league of the College of Sports Science of Nantong University, and the results confirmed the accuracy of the injury prediction of college basketball players based on FMS scores. It is hoped that this study can provide some reference for the research of injury prevention of basketball players.

We introduce a unique inductive bias to improve the reconstruction quality of Neural Radiance Fields (NeRF), NeRF employs the Fourier transform to map 3D coordinates to a high-dimensional space, enhancing the representation of high-frequency information in scenes. However, this transformation often introduces significant noise, affecting NeRF's robustness. Our approach allocates attention effectively by segregating channels within NeRF using attention-based gating units. We conducted experiments on an open-source data set to demonstrate the effectiveness of our method, which leads to significant improvements in the quality of synthesised new-view images compared to state-of-the-art methods. Notably, we achieve an average PSNR increase of 0.17 compared to the original NeRF. Furthermore, our method is implemented through a carefully designed special Multi-Layer Perceptron (MLP) architecture, ensuring compatibility with most existing NeRF-based methods.

Food quality is defined as a collection of properties that differentiate each unit and influences acceptability degree of food by users or consumers. Owing to the nature of food, food quality prediction is highly significant after specific periods of storage or before use by consumers. However, the accuracy is the major problem in the existing methods. Hence, this paper presents a BEFA_DRNN approach for accurate food quality prediction using dairy products. Firstly, input data is fed to data normalisation phase, which is performed by min-max normalisation. Thereafter, normalised data is given to feature fusion phase that is conducted employing DNN with Canberra distance. Then, fused data is subjected to data augmentation stage, which is carried out utilising oversampling technique. Finally, food quality prediction is done wherein milk is graded employing DRNN. The training of DRNN is executed by proposed BEFA that is a combination of BES and FA. Additionally, BEFA_DRNN obtained maximum accuracy, TPR and TNR values of 93.6%, 92.5% and 90.7%.

5G mobile communication systems are an in-depth fusion of multi-radio access technologies characterised with frequent handover between cells. Handover management is a particularly challenging issue for 5G networks development. In this article, a novel optimised handover framework is proposed to find the optimal network to connect with a good quality of service in accordance with the user's preferences. This framework is based on an extension of IEEE 802.21 standard with new components and new service primitives for seamless handover. Moreover, the proposed vertical handover process is based on an adaptive heuristic model aimed at achieving an optimised network during the decision-making stage. Simulation results demonstrate that, compared to other existing works, the proposed framework is capable of selecting the best network candidate accurately based on the quality-of-service requirements of the application, network conditions, mobile terminal conditions and user preferences. It significantly reduces the handover delay, handover blocking probability and packet loss rate.

Along with amazing advancements in the field of digital payments, this article seeks to provide a summary of research undertaken over the last four decades and to suggest areas in need of additional study. This study employs a two-pronged technique for analysing its data. The first is concerned with performance analysis, and the second with science mapping. The study uses the apps VOS viewer and R-studio to do bibliometric data analysis. From 1982 until May 2022, the most trustworthy database, Scopus, is used to compile a database of 923 publications The findings of this study identify the scope of current research interest, which is explored with critical contributions from a variety of authors, journals, countries, affiliations, keyword analysis, citation analysis, co-citation analysis, and bibliometric coupling, as well as a potential research direction for further investigation in this emerging field.

The banking sector has experienced a substantial technological shift that has opened up new and better opportunities for its customers. Based on their technological expenditures, the study assessed the two biggest public Indian banks and the two biggest private Indian banks. The most crucial statistical techniques used to demonstrate the aims are realistic are bivariate correlations and ordinary least squares. This work aims to establish a connection between research and a technology index that serves as a proxy for operational efficiency. The results show that for both public and private banks, the technology index positively influences operational efficiency metrics like IT costs, marketing costs, and compensation costs. This suggests that when the technology index increases, so do IT, marketing, and compensation costs, even though it has been shown that the technology index favourably improves operational efficiency measures like depreciation and printing. This means that the cost to banks is high despite greater investment in technology for these activities.

In order to further improve the public health assurance system and the infectious diseases early warning system to give play to their positive roles and enhance their collaborative capacity, this paper, based on the big and thick data analytics technology, designs a 'rolling-type' data synergy model. This model covers districts and counties, municipalities, provinces, and the country. It forms a data blockchain for the public health assurance system and enables high sharing of data from existing system platforms such as the infectious diseases early warning system, the hospital medical record management system, the public health data management system, and the health big and thick data management system. Additionally, it realises prevention, control and early warning by utilising data mining and synergy technologies, and ideally solves problems of traditional public health assurance system platforms such as excessive pressure on the 'central node', poor data tamper-proofing capacity, low transmission efficiency of big and thick data, bad timeliness of emergency response, and so on. The realisation of this technology can greatly improve the application and analytics of big and thick data and further enhance the public health assurance capacity.

The utilisation of big data presents significant opportunities for businesses to create value and gain a competitive edge. This capability enables firms to anticipate and uncover information quickly and intelligently. The author introduces a human resource scheduling optimisation strategy using a parallel network fusion structure model. The author's approach involves designing a set of network structures based on parallel networks and streaming media, enabling the macro implementation of the enterprise parallel network fusion structure. Furthermore, the author proposes a human resource scheduling optimisation method based on a parallel deep learning network fusion structure. It combines convolutional neural networks and transformer networks to fuse streaming media features, thereby achieving comprehensive identification of the effectiveness of the current human resource scheduling in enterprises. The result shows that the macro and deep learning methods achieve a recognition rate of 87.53%, making it feasible to assess the current state of human resource scheduling in enterprises.

Emergency disaster management presents substantial risks and obstacles to shield tunnel building projects, particularly in the event of water leakage accidents. Contemporary water leak detection is critical for guaranteeing safety by reducing the likelihood of disasters and the severity of any resulting damages. However, it can be difficult. Deep learning models can analyse images taken inside the tunnel to look for signs of water damage. This study introduces a unique strategy that employs deep learning techniques, generative adversarial networks (GAN) with long short-term memory (LSTM) for water leakage detection i shield tunnel construction (WLD-STC) to conduct classification and prediction tasks on the massive image dataset. The results demonstrate that for identifying and analysing water leakage episodes during shield tunnel construction, the WLD-STC strategy using LSTM-based GAN networks outperformed other methods, particularly on huge data.

To achieve psychological and emotional analysis of massive internet chats, researchers have used statistical methods, machine learning, and neural networks to analyse the dynamic tendencies of texts dynamically. For long readers, the author first compares and explores the differences between the two psychoanalysis algorithms based on the emotion dictionary and machine learning for simple sentences, then studies the expansion algorithm of the emotion dictionary, and finally proposes an extended text psychoanalysis algorithm based on conditional random field. According to the experimental results, the mental dictionary's accuracy, recall, and F-score based on the cognitive understanding of each additional ten words were calculated. The optimisation decreased, and the memory and F-score improved. An <i>F</i>-value greater than 1, which is the most effective indicator for evaluating the effectiveness of a mental analysis problem, can better demonstrate that the algorithm is adaptive in the literature dictionary. It has been proven that this scheme can achieve good results in analysing emotional tendencies and has higher efficiency than ordinary weight-based psychological sentiment analysis algorithms.

In the context of smart city development, the managed pressure drilling (MPD) drilling process faces many uncertainties, but the characteristics of the process are complex and require accurate wellbore pressure control. However, this process runs the risk of introducing un-modelled dynamics into the system. To this problem, this paper employs neural network control techniques to construct a dual-network system for throttle pressure control, the design encompasses both the controller and identifier components. The radial basis function (RBF) network and proportional features are connected in parallel in the controller structure, and the RBF network learning algorithm is used to train the identifier structure. The simulation results show that the actual wellbore pressure can quickly track the reference pressure value when the pressure setpoint changes. In addition, the controller based on neural network realises effective control, which enables the system to track the input target quickly and achieve stable convergence.

In order to reduce the challenges encountered by learners and educators in engaging in educational activities, this paper classifies learners' roles in virtual learning communities, and explores the role of behaviour characteristics and their positions in collaborative knowledge construction networks in promoting the process of knowledge construction. This study begins with an analysis of the relationship structure among learners in the virtual learning community and then applies the FCM algorithm to arrange learners into various dimensional combinations and create distinct learning communities. The test results demonstrate that the FCM method performs consistently during the clustering process, with less performance oscillations, and good node aggregation, the ARI value of the model is up to 0.90. It is found that they play an important role in the social interaction of learners' virtual learning community, which plays a certain role in promoting the development of artificial intelligence.

To explore the computer-aided translation technology based on the intelligent algorithm of edge computing. This paper presents the research on computer-aided translation technology based on edge computing intelligent algorithm. In the K-means computer edge algorithm, it analyses the traditional way of average resource allocation and the way of virtual machine allocation. In the process of online solution, we have a more detailed understanding of the data information at the edge, and also avoid the connection relationship between network users and the platform, which has a certain impact on the internal operation efficiency of the system. The network user group is divided into several different types of existence through K-means computer algorithm, and various information resources are counted according to their own characteristics. Computer-aided translation technology can significantly improve the quality of translation, improve the translation efficiency, and reduce the translation cost.

As IOT develops drastically these years, the application of cloud computing in many fields has become possible. In this paper, we take low-voltage current transformers in power systems as the research object and propose a TCN-BI-GRU power measurement method that incorporates the signal characteristics based on the transformer input and output. Firstly, the basic signal enhancement extraction of input and output is completed by using EMD and correlation coefficients. Secondly, multi-dimensional feature extraction is completed to improve the data performance according to the established TCN network. Finally, the power prediction is completed by using BI-GRU, and the results show that the RMSE of this framework is 5.69 significantly lower than other methods. In the laboratory test, the device after being subjected to strong disturbance, its correlation coefficient feature has a large impact, leading to a large deviation in the prediction, which provides a new idea for future intelligent prediction.

With the development of urban construction, its spatial evolution is also influenced by behavioural actors such as enterprises, residents, and environmental factors, leading to some decision-making behaviours that are not conducive to urban public space and environmental design. At the same time, some cities are vulnerable to various factors such as distance factors, transportation factors, and human psychological factors during the construction of public areas, resulting in a decline in the quality of urban human settlements. Urban public space is the guarantee of urban life. For this, in order to standardise urban public space and improve the quality of urban living environment, the standardisation of the environment of urban public space is required. The rapid development of intelligent algorithms and fuzzy control provides technical support for the environmental design of urban public spaces. Through the modelling of intelligent algorithms and the construction of fuzzy space, it can meet the diverse.

Physiological indicators are an important basis for reflecting the physiological health status of the human body and play an important role in medical practice. Association rules have also been one of the important research hotspots in recent years. This study aims to create a data mining system of association rules and artificial intelligence in biochemical indicators of sports training. This article uses Markov logic for network creation and system training, and tests whether the Markov logic network can be associated with the training system. The results show that the accuracy and recall rate obtained are about 90%, which shows that it is feasible to establish biochemical indicators of sports training based on Markov logic network, and the system has universal, guiding and constructive significance, ensuring that the construction of training system indicators will not go in the wrong direction.