Search results

Recent years have witnessed an unexpected and astonishing rise of AI-generated (AIGC), thanks to the rapid advancement of technology and the omnipresence of social media. AIGCs created to mislead are more commonly known as DeepFakes, which erode our trust in online information and have already caused real damage. Thus, countermeasures must be developed to limit the negative impacts of AIGC. This position paper aims to provide a conceptual analysis of the impact of DeepFakes considering the production cost and overview counter technologies to fight DeepFakes. We will also discuss future perspectives of AIGC and their counter technology.

We summarize recent developments in generative AI and AIGC, as well as technical developments to mitigate the harmful impacts of DeepFakes. We also provide an analysis of the cost-effect tradeoff of DeepFakes.

The mitigation of DeepFakes call for multi-disciplinary research across the traditional disciplinary boundaries.

Government and business sectors need to work together to provide sustainable solutions to the DeepFake problem.

The research and development in counter-technologies and other mitigation measures of DeepFakes are important components for the health of future information ecosystem and democracy.

Unlike existing reviews in this topic, our position paper focuses on the insights and perspective of this vexing sociotechnical problem of our time, providing a more global picture of the solutions landscape.

Classification of remote sensing images (RSI) is a challenging task in computer vision. Recently, researchers have proposed a variety of creative methods for automatic recognition of RSI, and feature fusion is a research hotspot for its great potential to boost performance. However, RSI has a unique imaging condition and cluttered scenes with complicated backgrounds. This larger difference from nature images has made the previous feature fusion methods present insignificant performance improvements.

This work proposed a two-convolutional neural network (CNN) fusion method named main and branch CNN fusion network (MBC-Net) as an improved solution for classifying RSI. In detail, the MBC-Net employs an EfficientNet-B3 as its main CNN stream and an EfficientNet-B0 as a branch, named MC-B3 and BC-B0, respectively. In particular, MBC-Net includes a long-range derivation (LRD) module, which is specially designed to learn the dependence of different features. Meanwhile, MBC-Net also uses some unique ideas to tackle the problems coming from the two-CNN fusion and the inherent nature of RSI.

Extensive experiments on three RSI sets prove that MBC-Net outperforms the other 38 state-of-the-art (STOA) methods published from 2020 to 2023, with a noticeable increase in overall accuracy (OA) values. MBC-Net not only presents a 0.7% increased OA value on the most confusing NWPU set but also has 62% fewer parameters compared to the leading approach that ranks first in the literature.

MBC-Net is a more effective and efficient feature fusion approach compared to other STOA methods in the literature. Given the visualizations of grad class activation mapping (Grad-CAM), it reveals that MBC-Net can learn the long-range dependence of features that a single CNN cannot. Based on the tendency stochastic neighbor embedding (t-SNE) results, it demonstrates that the feature representation of MBC-Net is more effective than other methods. In addition, the ablation tests indicate that MBC-Net is effective and efficient for fusing features from two CNNs.

The paper aims to tackle the classification of Remote Sensing Images (RSIs), which presents a significant challenge for computer algorithms due to the inherent characteristics of clustered ground objects and noisy backgrounds. Recent research typically leverages larger volume models to achieve advanced performance. However, the operating environments of remote sensing commonly cannot provide unconstrained computational and storage resources. It requires lightweight algorithms with exceptional generalization capabilities.

This study introduces an efficient knowledge distillation (KD) method to build a lightweight yet precise convolutional neural network (CNN) classifier. This method also aims to substantially decrease the training time expenses commonly linked with traditional KD techniques. This approach entails extensive alterations to both the model training framework and the distillation process, each tailored to the unique characteristics of RSIs. In particular, this study establishes a robust ensemble teacher by independently training two CNN models using a customized, efficient training algorithm. Following this, this study modifies a KD loss function to mitigate the suppression of non-target category predictions, which are essential for capturing the inter- and intra-similarity of RSIs.

This study validated the student model, termed KD-enhanced network (KDE-Net), obtained through the KD process on three benchmark RSI data sets. The KDE-Net surpasses 42 other state-of-the-art methods in the literature published from 2020 to 2023. Compared to the top-ranked method’s performance on the challenging NWPU45 data set, KDE-Net demonstrated a noticeable 0.4% increase in overall accuracy with a significant 88% reduction in parameters. Meanwhile, this study’s reformed KD framework significantly enhances the knowledge transfer speed by at least three times.

This study illustrates that the logit-based KD technique can effectively develop lightweight CNN classifiers for RSI classification without substantial sacrifices in computation and storage costs. Compared to neural architecture search or other methods aiming to provide lightweight solutions, this study’s KDE-Net, based on the inherent characteristics of RSIs, is currently more efficient in constructing accurate yet lightweight classifiers for RSI classification.

This study aims to detect consumer sarcasm through inconsistencies in sentiment features between text and images of hotel reviews.

This paper proposes a model for sarcasm detection based on multimodal deep learning using reviews of three hotel brands collected from two travel platforms, which can identify emotional inconsistencies within a modality and across modalities. Text-image interaction information is explored using graph neural networks (GNN) to detect essential clues in sarcasm sentiment.

The research results show that the multimodal deep learning model outperforms other baseline models, which can help to understand hotel service evaluation and provide hotel managers with decision-making opinions.

This research can help hoteliers in two ways: detecting service quality and formulating strategies. By selecting reference hotel brands, hoteliers can better assess their level of service quality (optimal resource allocation ensues); therefore, sarcasm detection research is not only beneficial for hotel managers seeking to improve service quality. The multimodal deep learning method introduced in the present study can be replicated in other industries to help travel platforms optimize their products and services.

本研究通过分析酒店评论文本和图像之间情感特征的不一致性来检测消费者的讽刺。

本文提出了一种基于多模态深度学习的讽刺检测模型, 使用从两个旅行平台收集的三个酒店品牌的评论, 该模型能够识别模态内部和模态之间的情感不一致性。利用图神经网络（GNN）探索文本-图像交互信息, 以检测讽刺情感中的关键线索。

研究结果显示, 多模态深度学习模型优于其他基线模型, 这有助于理解酒店服务评估, 并为酒店经理提供决策建议。

该研究可以在两方面帮助酒店业者：检测服务质量和制定策略。通过选择参考酒店品牌, 酒店业者可以更好地评估其服务质量水平（随之而来的是最佳资源分配）, 因此, 讽刺检测研究不仅有助于寻求提高服务质量的酒店经理。本研究介绍的多模态深度学习方法可以在其他行业复制, 帮助旅行平台优化其产品和服务。

This study aims to examine how dynamic pricing impacts customer perceptions of restaurants and sentiment toward prices via online reputation metrics. In addition, to deepen the debate on dynamic pricing, a novel definition is drawn by exploring the specific forms of discrimination that can manifest in different industries.

Leveraging a comprehensive data set of restaurant reviews sourced from TripAdvisor, the study focuses on restaurants affiliated with one of the largest groups of restaurants in Spain. We used a quasi-experimental method (difference-in-differences), to study how dynamic pricing strategies influence customers’ perceptions of value based on numerical ratings. Meanwhile, we used a Bidirectional Encoder Representations from Transformers model on the textual component of reviews to dissect the emotional nuances of dynamic pricing.

Results did not reveal a causal impact of dynamic pricing strategies on customers’ perceptions. Moreover, the sentiment analysis shows no heightened negative view after introducing dynamic pricing in restaurants compared to the control group. Contrary to what previous literature suggests, our findings indicate that implementing dynamic pricing does not adversely affect customers’ perceptions or sentiments regarding prices in restaurants.

The quasi-experimental setting of the study presents inherent challenges in establishing causality that require further investigation using controlled experimental settings. Nevertheless, our study reveals that restaurant customers do not perceive dynamic pricing as unfair. This finding is critical for restaurant managers when considering the implementation of dynamic pricing and revenue management strategies. In addition, our study highlights the importance of considering not only numerical ratings but customer sentiment analysis as well. This more holistic approach to assessing the impact of pricing strategies can give restaurant managers a deeper understanding of customer reactions. In addition, a more rigorous definition of dynamic pricing is provided, clarifying its nature and its distinction in using different price discrimination.

This study contributes to the evolving understanding of dynamic pricing strategies’ impact on customers’ perceptions and sentiments in the restaurant industry. It aims to fill the gap in understanding customer reactions to algorithmically determined prices (via revenue management systems such as DynamEat) in this industry. The combination of causal inference and sentiment analysis offers a novel perspective, shedding light on the nuanced connections between dynamic pricing implementation and customers’ emotions.

本研究考察动态定价如何通过在线声誉指标影响顾客对餐厅的感知和对价格的情绪。此外, 为了深化对动态定价的讨论, 通过探索不同行业中可能表现出的具体歧视形式, 提出了一个新的定义。

利用从TripAdvisor获取的餐厅评论的全面数据集, 研究聚焦于与西班牙最大的餐厅集团之一相关联的餐厅。我们采用了准实验方法（差异中的差异）, 研究动态定价策略如何根据数值评分影响顾客对价值的感知。同时, 我们运用BERT模型对评论的文本成分进行分析, 以解析动态定价的情感细微差别。

结果没有揭示动态定价策略对顾客感知产生因果影响。此外, 情绪分析显示, 在餐厅引入动态定价后, 与对照组相比, 没有增加消极观点。与以往文献所述相反, 我们的发现表明, 实施动态定价并不会对顾客对价格的感知或情绪产生负面影响。

研究的准实验设置存在确立因果关系的固有挑战, 需要通过控制实验设置进一步调查。尽管如此, 我们的研究揭示了餐厅顾客不认为动态定价不公平。这一发现对餐厅经理在考虑实施动态定价和收入管理策略时至关重要。此外, 我们的研究强调, 考虑顾客情绪分析和数值评分的重要性。这种更全面的方法评估定价策略的影响, 可以让餐厅经理更深入地理解顾客反应。此外, 提供了一个更严格的动态定价定义, 澄清了其性质及其在使用不同价格歧视中的区别。

本研究对于理解动态定价策略对餐厅行业顾客感知和情绪影响的不断发展有所贡献。它旨在填补对客户对算法确定的价格（通过收入管理系统（RMS）例如DynamEat）在此行业中反应的理解空白。因果推断与情绪分析的结合提供了新的视角, 揭示了动态定价实施与顾客情绪之间微妙的联系。

Este estudio examina cómo la fijación dinámica de precios impacta en las percepciones de los clientes de los restaurantes y en el sentimiento hacia los precios a través de métricas de reputación en línea. Además, para profundizar en el debate sobre la fijación dinámica de precios, se propone una definición novedosa explorando las formas específicas de discriminación que pueden manifestarse en diferentes industrias.

Utilizando un conjunto de datos exhaustivo de reseñas de restaurantes obtenidas de TripAdvisor, el estudio se centra en los restaurantes afiliados a uno de los mayores grupos de restaurantes en España. Empleamos un método cuasiexperimental (diferencias en diferencias) para estudiar cómo las estrategias de precios dinámicos influyen en las percepciones de valor de los clientes basándonos en las calificaciones numéricas. Mientras tanto, empleamos un modelo BERT en el componente textual de las reseñas para desentrañar los matices emocionales de la fijación dinámica de precios.

Los resultados no revelaron un impacto causal de las estrategias de precios dinámicos en las percepciones de los clientes. Además, el análisis de sentimiento no muestra una visión negativa aumentada después de introducir la fijación dinámica de precios en los restaurantes en comparación con el grupo de control. Contrariamente a lo que sugiere la literatura previa, nuestros hallazgos indican que la implementación de precios dinámicos no afecta negativamente las percepciones o los sentimientos de los clientes respecto a los precios en los restaurantes.

La configuración cuasiexperimental del estudio presenta desafíos inherentes para establecer la causalidad que requieren una investigación más profunda utilizando entornos experimentales controlados. Sin embargo, nuestro estudio revela que los clientes de restaurantes no perciben la fijación de precios dinámica como injusta. Este hallazgo es crítico para los gerentes de restaurantes al considerar la implementación de la fijación de precios dinámica y estrategias de gestión de ingresos. Además, nuestro estudio resalta la importancia de considerar no solo las calificaciones numéricas sino también el análisis del sentimiento del cliente. Este enfoque más holístico para evaluar el impacto de las estrategias de precios puede dar a los gerentes de restaurantes una comprensión más profunda de las reacciones de los clientes. Además, se proporciona una definición de fijación de precios dinámica más rigurosa, aclarando su naturaleza y su distinción en el uso de diferentes discriminaciones de precios.

Este estudio contribuye a la comprensión en evolución del impacto de las estrategias de fijación de precios dinámicos en las percepciones y sentimientos de los clientes en la industria restaurantera. Su objetivo es llenar el vacío en la comprensión de las reacciones de los clientes a los precios determinados algorítmicamente (a través de sistemas de gestión de ingresos (RMS) como DynamEat) en esta industria. La combinación de inferencia causal y análisis de sentimientos ofrece una perspectiva novedosa, arrojando luz sobre las conexiones matizadas entre la implementación de la fijación de precios dinámicos y las emociones de los clientes.

This study aims to analyse the literature published on ChatGPT and generative pre-trained transformer (GPT) available through Scopus to identify the top countries, institutions, authors, journals and titles in terms of publishing and citations in the area.The research also intends to determine the collaborative trends among countries and authors as well as top-used keywords on the topic identified from the analysed literature.

The researchers searched the Scopus database to collect and assess the literature on the topic. The paper used six applications such as Biblioshiny, VosViewer, Python, MS Access and Excel and Endnote to collect and analyse the literature.

It is found that European countries such as the USA, the UK and Germany took the lead in terms of publishing and impact in the area but the USA stood first with 90 publications and 1,720 citations in this connection. Likewise, the organization “Rheinisch-Westfälische Technische Hochschule Aachen” scored the top position regarding publishing, but Open AI remained on top for getting the highest citations (1,384). Apropos, the author “Choi, Y” produced the highest number of publications. The research also shares the collaborative patterns, top journals and publications in the area, as well as the top-used keywords on the topic.

To the best of the authors’ knowledge, this is the first study that shares a bibliometric analysis of literature published on GPT and ChatGPT. The research not only fills the research gap on the topic but also shares implications for relevant stakeholders and future research directions for future researchers.

This paper aims to facilitate the research and development of resilient navigation approaches, explore the robustness of adversarial training to different interferences and promote their practical applications in real complex environments.

In this paper, the authors first summarize the real accidents of self-driving cars and develop a set of methods to simulate challenging scenarios by introducing simulated disturbances and attacks into the input sensor data. Then a robust and transferable adversarial training approach is proposed to improve the performance and resilience of current navigation models, followed by a multi-modality fusion-based end-to-end navigation network to demonstrate real-world performance of the methods. In addition, an augmented self-driving simulator with designed evaluation metrics is built to evaluate navigation models.

Synthetical experiments in simulator demonstrate the robustness and transferability of the proposed adversarial training strategy. The simulation function flow can also be used for promoting any robust perception or navigation researches. Then a multi-modality fusion-based navigation framework is proposed as a light-weight model to evaluate the adversarial training method in real-world.

The adversarial training approach provides a transferable and robust enhancement for navigation models both in simulation and real-world.

In the COVID-19 era, sign language (SL) translation has gained attention in online learning, which evaluates the physical gestures of each student and bridges the communication gap between dysphonia and hearing people. The purpose of this paper is to devote the alignment between SL sequence and nature language sequence with high translation performance.

SL can be characterized as joint/bone location information in two-dimensional space over time, forming skeleton sequences. To encode joint, bone and their motion information, we propose a multistream hierarchy network (MHN) along with a vocab prediction network (VPN) and a joint network (JN) with the recurrent neural network transducer. The JN is used to concatenate the sequences encoded by the MHN and VPN and learn their sequence alignments.

We verify the effectiveness of the proposed approach and provide experimental results on three large-scale datasets, which show that translation accuracy is 94.96, 54.52, and 92.88 per cent, and the inference time is 18 and 1.7 times faster than listen-attend-spell network (LAS) and visual hierarchy to lexical sequence network (H2SNet) , respectively.

In this paper, we propose a novel framework that can fuse multimodal input (i.e. joint, bone and their motion stream) and align input streams with nature language. Moreover, the provided framework is improved by the different properties of MHN, VPN and JN. Experimental results on the three datasets demonstrate that our approaches outperform the state-of-the-art methods in terms of translation accuracy and speed.

Purpose: This chapter aims to unveil the challenges of adopting and using banking chatbots in India and identify the challenges of Chat Generative Pre-trained Transformer (ChatGPT) for future banking.

Need for the study: Unveiling the challenges of chatbots and ChatGPT in the banking industry in India is crucial to understand the limitations and areas of improvement to enhance customer experience, ensure data security, and maintain regulatory compliance.

Methodology: The researchers conducted a narrative review systematically summarising and analysing existing literature on chatbots and ChatGPT, providing a comprehensive overview of the challenges faced in the industry.

Findings: The authors identify perceived risk, platform quality, connectivity and infrastructure, data privacy and security, user education and acceptance, existing legacy systems, and regulatory guidelines as the challenges of adopting chatbots. Additionally, the findings reveal that the challenges posed by ChatGPT in future banking include the potential reduction of traditional banking jobs, linguistic diversity, data privacy and security, ethical considerations and bias mitigation, explainability and accountability, integration with existing banking systems, and user trust and acceptance. However, implementing these new technologies also presents opportunities for individuals with unique human skills, such as critical thinking, empathy, and creativity, which are difficult to replace with technology.

Practical implications: By minimising the challenges of ChatGPT and chatbots, the banking industry could achieve improved customer service, cost efficiency, automation of routine tasks, and 24/7 availability, leading to enhanced customer satisfaction and operational efficiency in the banking industry. Additionally, these artificial intelligence (AI) tools enable data-driven insights, personalised experiences, scalability, and efficient handling of large customer volumes, contributing to better decision-making and enhanced customer engagement.

This study aims to apply the appreciative inquiry approach (AI) to develop a tourism strategy for poverty alleviation in marginalised communities. The focus is to provide practical insights for leveraging tourism to drive positive socio-economic change for the impoverished, using Rosetta, a port city in Egypt with cultural and historical significance, as a case study.

This qualitative applied study uses the four-D phases of AI and thematic analysis to strategise tourism development in Rosetta. Through interviews, focus groups and field visits, the study identifies tourism potential, stakeholder aspirations and actionable strategies for sustainable development. The approach prioritises a bottom-up, community-centric and stakeholder-involved process, aiming for inclusive and equitable growth.

The study revealed Rosetta’s underutilised tourism potential, emphasising heritage tourism. Although tourism offers some economic benefits, its impact on alleviating poverty in Rosetta remains limited. A holistic strategy for tourism development in Rosetta is proposed for economic growth and poverty reduction, focusing on sustainable management, local empowerment, enhanced marketing, improved infrastructure and diversified tourism offerings.

While AI is not new in qualitative studies, the novelty of this study lies in its application to tourism planning for poverty alleviation in a marginalised community like Rosetta, introducing a comprehensive tourism strategy with an original framework applicable to comparable destinations. The study’s significance is emphasised by providing actionable strategies for policymakers, valuable insights for practitioners and enriching the discourse and methodology on pro-poor tourism for academics, representing a step towards filling the gap between theoretical concepts and practical strategies.