[go: up one dir, main page]
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
5.3
2.5
Journals
Applied Sciences
Special Issues
Holistic AI Technologies and Applications
applsci-logo
Submit to Special Issue Submit Abstract to Special Issue Review for Applied Sciences Propose a Special Issue

Journal Menu

► Journal Menu

Journal Browser

► Journal Browser
share announcement

Holistic AI Technologies and Applications

A special issue of Applied Sciences (ISSN 2076-3417). This special issue belongs to the section "Computing and Artificial Intelligence".

Deadline for manuscript submissions: 30 June 2025 | Viewed by 14245

Special Issue Editors

Dr. Junlan Feng

E-Mail Website
Guest Editor
China Mobile Research Institute, Beijing 100053, China
Interests: artificial intelligence; knowledge engineering; data mining; speech processing; natural language processing
Prof. Dr. Guilin Qi

E-Mail Website
Guest Editor
School of Computer Science and Engineering, Southeast University, Nanjing 211189, China
Interests: natural language processing; knowledge graph; multimodal learning
Special Issues, Collections and Topics in MDPI journals

Special Issue Information

Dear Colleagues,

Holistic artificial intelligence mainly studies the theories, technologies, mechanisms, paradigms and frameworks required for the systematic reconstruction of artificial intelligence. It comprises big loop AI (to realize an end-to-end optimization with cascade and parallel AI capabilities), atomized AI (to dismantle and refactor AI capability in a reusable manner), network native AI (to standardize AI capability and computability as a service for on-demand schedule via network), and trusted AI (to ensure traceability, trustworthiness, auditability and defensibility during AI process).  Relying on the ubiquitous network and computability, holistic artificial intelligence realizes flexible and efficient configuration, scheduling, training, and deployment of AI capabilities in an open environment, so as to meet the increasingly rich digital intelligent business needs while ensuring that AI business is trusted and controllable.

This Special Issue would like to highlight new and innovative work focused on holistic AI. We invite authors to present high-quality research work in one or more areas revolving around the current state of the art. This Special Issue intends to explore ‘Advances and Applications in Holistic Artificial Intelligence’, but is not restricted to big loop AI, atomized AI, network native AI, trusted AI, and its applications in specific domains like computer vision, natural language processing, speech, knowledge graph, data analysis, network intelligence, etc.

Dr. Junlan Feng
Prof. Dr. Guilin Qi
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Applied Sciences is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

  • big loop AI
  • atomized AI
  • network native AI
  • trusted AI
  • holistic AI in computer vision
  • holistic AI in knowledge engineering
  • holistic AI in natural language processing

Benefits of Publishing in a Special Issue

  • Ease of navigation: Grouping papers by topic helps scholars navigate broad scope journals more efficiently.
  • Greater discoverability: Special Issues support the reach and impact of scientific research. Articles in Special Issues are more discoverable and cited more frequently.
  • Expansion of research network: Special Issues facilitate connections among authors, fostering scientific collaborations.
  • External promotion: Articles in Special Issues are often promoted through the journal's social media, increasing their visibility.
  • e-Book format: Special Issues with more than 10 articles can be published as dedicated e-books, ensuring wide and rapid dissemination.

Further information on MDPI's Special Issue policies can be found here.

Published Papers (6 papers)

Download All Papers
Order results
Result details
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:

Research

35 pages, 2815 KiB  
Article
A Knowledge Graph Framework to Support Life Cycle Assessment for Sustainable Decision-Making
by Lucas Greif, Svenja Hauck, Andreas Kimmig and Jivka Ovtcharova
Appl. Sci. 2025, 15(1), 175; https://doi.org/10.3390/app15010175 - 28 Dec 2024
Viewed by 930
Abstract
This study introduces a comprehensive knowledge graph (KG)-based framework designed to support sustainable decision-making by integrating, enriching, and analyzing heterogeneous data sources. The proposed methodology leverages domain expertise, real-world data, and synthetic data generated through language models to address challenges in life cycle [...] Read more.
This study introduces a comprehensive knowledge graph (KG)-based framework designed to support sustainable decision-making by integrating, enriching, and analyzing heterogeneous data sources. The proposed methodology leverages domain expertise, real-world data, and synthetic data generated through language models to address challenges in life cycle assessment (LCA), particularly data scarcity and inconsistency. By modeling the entire product lifecycle, including engineering, production, usage, and disposal phases, the framework facilitates early-stage design decision-making and provides actionable insights for sustainability improvements. The methodology is validated through a case study on 3D printing (3DP), demonstrating its ability to manage complex data, highlight relationships between engineering decisions and environmental impacts, and mitigate data scarcity in the early phases of product development in the context of LCAs. In conclusion, the results demonstrate the framework’s potential to drive sustainable innovation in manufacturing. Full article
(This article belongs to the Special Issue Holistic AI Technologies and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>The four phases of life cycle assessment.</p> Full article ">Figure 2
<p>Methodology.</p> Full article ">Figure 3
<p>Stages of the product life cycle that are included in the methodology: from the engineering phase to the disposal phase.</p> Full article ">Figure 4
<p>General schema for LPG model.</p> Full article ">Figure 5
<p>Excerpt of adapted KG schema for 3DP process.</p> Full article ">Figure 6
<p>LPG model for 3DP process with an excerpt highlighting the different nodes and relations.</p> Full article ">
31 pages, 4387 KiB  
Article
Evaluating the Performance of Large Language Models in Predicting Diagnostics for Spanish Clinical Cases in Cardiology
by Julien Delaunay and Jordi Cusido
Appl. Sci. 2025, 15(1), 61; https://doi.org/10.3390/app15010061 - 25 Dec 2024
Cited by 1 | Viewed by 709
Abstract
This study explores the potential of large language models (LLMs) in predicting medical diagnoses from Spanish-language clinical case descriptions, offering an alternative to traditional machine learning (ML) and deep learning (DL) techniques. Unlike ML and DL models, which typically rely on extensive domain-specific [...] Read more.
This study explores the potential of large language models (LLMs) in predicting medical diagnoses from Spanish-language clinical case descriptions, offering an alternative to traditional machine learning (ML) and deep learning (DL) techniques. Unlike ML and DL models, which typically rely on extensive domain-specific training and complex data preprocessing, LLMs can process unstructured text data directly without the need for specialized training on medical datasets. This unique characteristic of LLMs allows for faster implementation and eliminates the risks associated with overfitting, which are common in ML and DL models that require tailored training for each new dataset. In this research, we investigate the capacities of several state-of-the-art LLMs in predicting medical diagnoses based on Spanish textual descriptions of clinical cases. We measured the impact of prompt techniques and temperatures on the quality of the diagnosis. Our results indicate that Gemini Pro and Mixtral 8x22b generally performed well across different temperatures and techniques, while Medichat Llama3 showed more variability, particularly with the few-shot prompting technique. Low temperatures and specific prompt techniques, such as zero-shot and Retrieval-Augmented Generation (RAG), tended to yield clearer and more accurate diagnoses. This study highlights the potential of LLMs as a disruptive alternative to traditional ML and DL approaches, offering a more efficient, scalable, and flexible solution for medical diagnostics, particularly in the non-English-speaking population. Full article
(This article belongs to the Special Issue Holistic AI Technologies and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Template of the evaluator prompt for diagnoses evaluation. This template is adapted to the metric used (‘&lt;metric&gt;’), the scale (‘&lt;scale&gt;’), and might comprise examples (‘&lt;examples&gt;’).</p> Full article ">Figure 2
<p>Precision score for each Diagnostics Model for each prompt technique with a low temperature.</p> Full article ">Figure 3
<p>Precision score for each Diagnostics Model with a high temperature for the three prompt techniques.</p> Full article ">Figure 4
<p>The recall evaluated the three Diagnostics Models with low temperature (0) and different prompt techniques.</p> Full article ">Figure 5
<p>The recall was obtained by each Diagnostics Model with high temperature (2) and zero-shot, few-shot, and RAG prompt.</p> Full article ">Figure 6
<p>Completeness score for each Diagnostics Model with a temperature of 0 and diverse prompt techniques.</p> Full article ">Figure 7
<p>Completeness score for each Diagnostics Model with a temperature of 2 and zero-shot, few-shot, and RAG prompts.</p> Full article ">Figure A1
<p>This prompt template is used to instruct the LLM to generate possible diagnoses based on patient information.</p> Full article ">Figure A2
<p>This prompt template is used to ask the LLM to list the most probable diagnoses based on a given clinical case and explain why each diagnosis is a possibility. Text in red is the emotional stimuli added, while text in blue might be replaced depending on the parameters evaluated (zero-shot, few-shot, RAG).</p> Full article ">Figure A3
<p>A scale from 1 to 5 was used to evaluate the precision of the predicted diagnoses compared to the true diagnosis.</p> Full article ">Figure A4
<p>This prompt is used to evaluate the recall of the predicted diagnoses compared to the true diagnosis on a scale from 1 to 5.</p> Full article ">Figure A5
<p>This scale is used to help an LLM evaluate the completeness of diagnoses predicted by a model compared to the true diagnosis.</p> Full article ">
16 pages, 774 KiB  
Article
CTGGAN: Controllable Text Generation with Generative Adversarial Network
by Zhe Yang, Yi Huang, Yaqin Chen, Xiaoting Wu, Junlan Feng and Chao Deng
Appl. Sci. 2024, 14(7), 3106; https://doi.org/10.3390/app14073106 - 8 Apr 2024
Viewed by 1995
Abstract
Controllable Text Generation (CTG) aims to modify the output of a Language Model (LM) to meet specific constraints. For example, in a customer service conversation, responses from the agent should ideally be soothing and address the user’s dissatisfaction or complaints. This imposes significant [...] Read more.
Controllable Text Generation (CTG) aims to modify the output of a Language Model (LM) to meet specific constraints. For example, in a customer service conversation, responses from the agent should ideally be soothing and address the user’s dissatisfaction or complaints. This imposes significant demands on controlling language model output. However, demerits exist among traditional methods. Promoting and fine-tuning language models exhibit the “hallucination” phenomenon and cannot guarantee complete adherence to constraints. Conditional language models (CLM), which map control codes into LM representations or latent space, require training the modified language models from scratch and a high amount of customized dataset is demanded. Decoding-time methods employ Bayesian Rules to modify the output of the LM or model constraints as a combination of energy functions and update the output along the low-energy direction. Both methods are confronted with the efficiency sampling problem. Moreover, there are no methods that consider the relation between constraints weights and the contexts, as is essential in actual applications such as customer service scenarios. To alleviate the problems mentioned above, we propose Controllable Text Generation with Generative Adversarial Networks (CTGGAN), which utilizes a language model with logits bias as the Generator to produce constrained text and employs the Discriminator with learnable constraint weight combinations to score and update the generation. We evaluate the method in the text completion task and Chinese customer service dialogues scenario, and our method shows superior performance in metrics such as PPL and Dist-3. In addition, CTGGAN also exhibits efficient decoding compared to other methods. Full article
(This article belongs to the Special Issue Holistic AI Technologies and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>The overall framework of our model. We use the original LM encoder (such as GPT-2) to encode the text, leaving the original logits part unchanged (the dusty blue part in top-left of the picture), and adding the logits bias network (the orange part in top-left of the picture) as a modification to the original LM. The above network forms a complete Generator, which is used to generate text that satisfies the constraints. The right part is the Discriminator network, which is composed of multiple discriminators responsible for evaluating the text. The weight of each discriminator is represented by the text encoding and the discriminator embedding, determined by the similarity.</p> Full article ">Figure 2
<p>Ablation results in 15 prefixes evaluation. W/O.DIS means generating texts with only Generator training.</p> Full article ">
20 pages, 5164 KiB  
Article
Exploring the Relationship between the Coverage of AI in WIRED Magazine and Public Opinion Using Sentiment Analysis
by Flavio Moriniello, Ana Martí-Testón, Adolfo Muñoz, Daniel Silva Jasaui, Luis Gracia and J. Ernesto Solanes
Appl. Sci. 2024, 14(5), 1994; https://doi.org/10.3390/app14051994 - 28 Feb 2024
Cited by 2 | Viewed by 2266
Abstract
The presence and significance of artificial intelligence (AI) technology in society have been steadily increasing since 2000. While its potential benefits are widely acknowledged, concerns about its impact on society, the economy, and ethics have also been raised. Consequently, artificial intelligence has garnered [...] Read more.
The presence and significance of artificial intelligence (AI) technology in society have been steadily increasing since 2000. While its potential benefits are widely acknowledged, concerns about its impact on society, the economy, and ethics have also been raised. Consequently, artificial intelligence has garnered widespread attention in news media and popular culture. As mass media plays a pivotal role in shaping public perception, it is crucial to evaluate opinions expressed in these outlets. Understanding the public’s perception of artificial intelligence is essential for effective public policy and decision making. This paper presents the results of a sentiment analysis study conducted on WIRED magazine’s coverage of artificial intelligence between January 2018 and April 2023. The objective of the study is to assess the prevailing opinions towards artificial intelligence in articles from WIRED magazine, which is widely recognized as one of the most reputable and influential publications in the field of technology and innovation. Using two sentiment analysis techniques, AFINN and VADER, a total of 4265 articles were analyzed for positive, negative, and neutral sentiments. Additionally, a term frequency analysis was conducted to categorize articles based on the frequency of mentions of artificial intelligence. Finally, a linear regression analysis of the mean positive and negative sentiments was performed to examine trends for each month over a five-year period. The results revealed a leading pattern: there was a predominant positive sentiment with an upward trend in both positive and negative sentiments. This polarization of sentiment suggests a shift towards more extreme positions, which should influence public policy and decision making in the near future. Full article
(This article belongs to the Special Issue Holistic AI Technologies and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Interest over time in “artificial intelligence” obtained from Google Trends.</p> Full article ">Figure 2
<p>Example of score of articles with both techniques.</p> Full article ">Figure 3
<p>Distribution of articles by category of mention of artificial intelligence (noMention = 0 times, LowMention = 1 time, Mention = from 2 to 4 times, stroMention = more than 4 times).</p> Full article ">Figure 4
<p>Distribution of overall sentiment in <span class="html-italic">WIRED</span> magazine articles on artificial intelligence, obtained using AFINN and VADER techniques.</p> Full article ">Figure 5
<p>Distribution of overall sentiment per category of mention in <span class="html-italic">WIRED</span> magazine articles on artificial intelligence, obtained using AFINN and VADER techniques.</p> Full article ">Figure 6
<p>Mean positive and negative score per month from January 2018 to April 2023 (64 months). The scale ranges of VADER [−4, +4] and AFINN [−5, +5] were adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 7
<p>AFINN mean positive and negative score per month within the category “noMention” from January 2018 to April 2023 (64 months). The scale ranges of AFINN [−5, +5] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 8
<p>AFINN mean positive and negative score per month within the category “LowMention” from January 2018 to April 2023 (64 months). The scale range of AFINN [−5, +5] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 9
<p>AFINN mean positive and negative score per month within the category “Mention” from January 2018 to April 2023 (64 months). The scale range of AFINN [−5, +5] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 10
<p>AFINN mean positive and negative score per month within the category “stroMention” from January 2018 to April 2023 (64 months). The scale range of AFINN [−5, +5] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 11
<p>VADER mean positive and negative score per month within the category “NoMention” from January 2018 to April 2023 (64 months). The scale range of VADER [−4, +4] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 12
<p>VADER mean positive and negative score per month within the category “lowMention” from January 2018 to April 2023 (64 months). The scale range of VADER [−4, +4] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 13
<p>VADER mean positive and negative score per month within the category “Mention” from January 2018 to April 2023 (64 months). The scale range of VADER [−4, +4] was adjusted to a unified scale of [−1, +1].</p> Full article ">Figure 14
<p>VADER mean positive and negative score per month within the category “stroMention” from January 2018 to April 2023 (64 months). The scale range of VADER [−4, +4] was adjusted to a unified scale of [−1, +1].</p> Full article ">
13 pages, 1663 KiB  
Article
EduChat: An AI-Based Chatbot for University-Related Information Using a Hybrid Approach
by Hoa Dinh and Thien Khai Tran
Appl. Sci. 2023, 13(22), 12446; https://doi.org/10.3390/app132212446 - 17 Nov 2023
Cited by 5 | Viewed by 5809
Abstract
The digital transformation has created an environment that fosters the development of effective chatbots. Through the fusion of artificial intelligence and data, these chatbots have the capability to provide automated services, optimize customer experiences, and reduce workloads for employees. These chatbots can offer [...] Read more.
The digital transformation has created an environment that fosters the development of effective chatbots. Through the fusion of artificial intelligence and data, these chatbots have the capability to provide automated services, optimize customer experiences, and reduce workloads for employees. These chatbots can offer 24/7 support, answer questions, perform transactions, and provide rapid information, contributing significantly to the sustainable development processes of businesses and organizations. ChatGPT has already been applied in various fields. However, to ensure that there is a chatbot providing accurate and useful information in a narrow domain, it is necessary to build, train, and fine-tune the model based on specific data. In this paper, we introduce EduChat, a chatbot system for university-related questions. EduChat is an effective artificial intelligence application designed by combining rule-based methods, an innovative improved random forest machine learning approach, and ChatGPT to automatically answer common questions related to universities, academic programs, admission procedures, student life, and other related topics. This chatbot system helps provide quick and easy information to users, thereby reducing the time spent searching for information directly from source documents or contacting support staff. The experiments have yielded positive results. Full article
(This article belongs to the Special Issue Holistic AI Technologies and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>A conversation example with EduChat. (User: Hi EduChat: Hello! I am an AI-based support staff member at the university. How can I assist you? User: Can I ask if the schedule for semester 1 is available yet? EduChat: You can access the portal at the following link: <a href="https://portal.huflit.edu.vn/" target="_blank">https://portal.huflit.edu.vn/</a> (accessed on 1 October 2023). Please log in and use the ‘View Schedule’ function. Make sure to select the correct academic year and semester you want to view. User: But I forgot my password. EduChat: You can click on ‘Forgot Password’ on the portal to reset your password.).</p> Full article ">Figure 2
<p>EduChat system architecture.</p> Full article ">Figure 3
<p>Classification results using machine learning and deep learning methods on the EduChat system.</p> Full article ">
14 pages, 2828 KiB  
Article
Reconstructed Prototype Network Combined with CDC-TAGCN for Few-Shot Action Recognition
by Aihua Wu and Songyu Ding
Appl. Sci. 2023, 13(20), 11199; https://doi.org/10.3390/app132011199 - 12 Oct 2023
Viewed by 1063
Abstract
Research on few-shot action recognition has received widespread attention recently. However, there are some blind spots in the current research: (1) The prevailing practice in many models is to assign uniform weights to all samples; nevertheless, such an approach may yield detrimental consequences [...] Read more.
Research on few-shot action recognition has received widespread attention recently. However, there are some blind spots in the current research: (1) The prevailing practice in many models is to assign uniform weights to all samples; nevertheless, such an approach may yield detrimental consequences for the model in the presence of high-noise samples. (2) Samples with similar features but different classes make it difficult for the model to be distinguished. (3) Skeleton data harbors rich temporal features, but most encoders face challenges in effectively extracting them. In response to these challenges, this study introduces a reconstructed prototype network (RC-PN) based on a prototype network framework and a novel spatiotemporal encoder. The RC-PN comprises two enhanced modules: Sample coefficient reconstruction (SCR) and a reconstruction loss function (LRC). SCR leverages cosine similarity between samples to reassign sample weights, thereby generating prototypes robust to noise interference and more adept at conveying conceptual essence. Simultaneously, the introduction of LRC enhances the feature similarity among samples of the same class while increasing feature distinctiveness between different classes. In the encoder aspect, this study introduces a novel spatiotemporal convolutional encoder called CDC-TAGCN. The temporal convolution operator is redefined in CDC-TAGCN. The vanilla temporal convolution operator can only capture the surface-level characteristics of action samples. Drawing inspiration from differential convolution (CDC), this research enhances TCN to CDC-TGCN. CDC-TGCN allows for the fusion of discrepant features from action samples into the features extracted by the vanilla convolutional operator. Extensive feasibility and ablation experiments are performed on the skeleton action dataset NTU-RGB + D 120 and Kinetics and compared with recent research. Full article
(This article belongs to the Special Issue Holistic AI Technologies and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Different colors represent different classes, and the dotted line represents the distance between the query sample and prototypes. According to Equation (4), the closer to a prototype, the higher the probability of belonging to that class.</p> Full article ">Figure 2
<p>In contrast to <a href="#applsci-13-11199-f001" class="html-fig">Figure 1</a>, this figure illustrates how SCR enables query samples that were originally misclassified into the blue class to be correctly classified into the green class.</p> Full article ">Figure 3
<p>The visualization illustrates the effect of <math display="inline"><semantics> <mrow> <msub> <mrow> <mi>L</mi> </mrow> <mrow> <mi>R</mi> <mi>C</mi> </mrow> </msub> </mrow> </semantics></math>, as observed from the results on the right. <math display="inline"><semantics> <mrow> <msub> <mrow> <mi>L</mi> </mrow> <mrow> <mi>R</mi> <mi>C</mi> </mrow> </msub> </mrow> </semantics></math> enhances the dispersion between classes while bringing samples of the same class closer together.</p> Full article ">Figure 4
<p>This figure takes a 5-way 5-shot task as an example to show the process of the prototype network in the training phase. The content of the encoder will be introduced in detail in the next section.</p> Full article ">Figure 5
<p>This is the framework diagram of the encoder, which consists of a total of 6 layers. The <math display="inline"><semantics> <mrow> <mn>1</mn> <mo>×</mo> <mn>1</mn> </mrow> </semantics></math> Conv represents the residual connections used between layers to improve model fitting. The features output by the encoder are then input into RC-PN.</p> Full article ">Figure 6
<p>From the figure, it can be seen that after the model converges, <math display="inline"><semantics> <mrow> <msub> <mrow> <mi>L</mi> </mrow> <mrow> <mi>R</mi> <mi>C</mi> </mrow> </msub> </mrow> </semantics></math> approximately accounts for about one tenth of the total loss.</p> Full article ">Figure 7
<p>Model performance as a function of <span class="html-italic">θ</span>.</p> Full article ">
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying articles 1-6
Back to TopTop