Pub Date : 2024-02-07DOI: 10.1109/ICAIC60265.2024.10433804
Trishla Shah, Raghav V. Sampangi, Angela Siegel
Over the years, mobile applications have brought about transformative changes in user interactions with digital services. Many of these apps however, are free and offer convenience at the cost of exchanging personal data. This convenience, however, comes with inherent risks to user privacy and security. This paper introduces a comprehensive methodology that evaluates the risks associated with sharing sensitive data through mobile applications. Building upon the Hierarchical Weighted Risk Scoring Model (HWRSM), this paper proposes an evaluation methodology for HWRSM, keeping in mind the implications of such risk scoring on real-world security scenarios. The methodology employs innovative risk scoring, considering various factors to assess potential security vulnerabilities related to sensitive terms. Practical assessments involving diverse set of Android applications, particularly in data-intensive categories, reveal insights into data privacy practices, vulnerabilities, and alignment with HWRSM scores. By offering insights into testing, validation, real-world findings, and model effectiveness, the paper aims to provide practical considerations to mobile application security discussions, facilitating informed approaches to address security and privacy concerns.
{"title":"Risk-Aware Mobile App Security Testing: Safeguarding Sensitive User Inputs","authors":"Trishla Shah, Raghav V. Sampangi, Angela Siegel","doi":"10.1109/ICAIC60265.2024.10433804","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433804","url":null,"abstract":"Over the years, mobile applications have brought about transformative changes in user interactions with digital services. Many of these apps however, are free and offer convenience at the cost of exchanging personal data. This convenience, however, comes with inherent risks to user privacy and security. This paper introduces a comprehensive methodology that evaluates the risks associated with sharing sensitive data through mobile applications. Building upon the Hierarchical Weighted Risk Scoring Model (HWRSM), this paper proposes an evaluation methodology for HWRSM, keeping in mind the implications of such risk scoring on real-world security scenarios. The methodology employs innovative risk scoring, considering various factors to assess potential security vulnerabilities related to sensitive terms. Practical assessments involving diverse set of Android applications, particularly in data-intensive categories, reveal insights into data privacy practices, vulnerabilities, and alignment with HWRSM scores. By offering insights into testing, validation, real-world findings, and model effectiveness, the paper aims to provide practical considerations to mobile application security discussions, facilitating informed approaches to address security and privacy concerns.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"17 5","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139893354","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-07DOI: 10.1109/ICAIC60265.2024.10433828
Trishla Shah, Raghav V. Sampangi, Angela Siegel
This research paper introduces a Hierarchical Weighted Risk Scoring Model specifically designed to assess the risk levels of mobile applications based on user inputs. Through an extensive review of literature on risk score calculation models and term sensitivity identification techniques, this study categorizes terms based on their sensitivity, particularly in relation to sensitive user inputs that may potentially lead to data leaks. The sensitivity of user terms are defined based on the guidelines from PIPEDA. By integrating these terms, along with test outcomes and weights, the model accurately calculates risk scores. The resulting risk assessments provide users with valuable insights, empowering them to make informed decisions and effectively manage risks associated with mobile application usage. This research contributes to the field by offering a user-centric framework that combines various risk score calculation models and term sensitivity identification techniques, tailored specifically for mobile applications and addressing the potential risks arising from sensitive user inputs.
{"title":"Mobile Application Security Risk Score: A sensitive user input-based approach","authors":"Trishla Shah, Raghav V. Sampangi, Angela Siegel","doi":"10.1109/ICAIC60265.2024.10433828","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433828","url":null,"abstract":"This research paper introduces a Hierarchical Weighted Risk Scoring Model specifically designed to assess the risk levels of mobile applications based on user inputs. Through an extensive review of literature on risk score calculation models and term sensitivity identification techniques, this study categorizes terms based on their sensitivity, particularly in relation to sensitive user inputs that may potentially lead to data leaks. The sensitivity of user terms are defined based on the guidelines from PIPEDA. By integrating these terms, along with test outcomes and weights, the model accurately calculates risk scores. The resulting risk assessments provide users with valuable insights, empowering them to make informed decisions and effectively manage risks associated with mobile application usage. This research contributes to the field by offering a user-centric framework that combines various risk score calculation models and term sensitivity identification techniques, tailored specifically for mobile applications and addressing the potential risks arising from sensitive user inputs.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"23 1","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139895495","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The growth of the Internet has aroused people’s attention toward network security. A secure network environment is fundamental for the expeditious and impeccable development of the Internet. The majority of internet-based tasks can be completed with the help of a web browser. Although many web applications add browser extensions to improve their functionality, some of these extensions are malicious and can access sensitive data without the user’s knowledge. Browser extensions with malicious intent present a growing security concern and have quickly become one of the most prevalent methods used to compromise Internet security. This is largely due to their widespread usage and the extensive privileges they possess. After being installed, these malicious extensions are executed and make an attempt to compromise the victim’s browser. This makes them particularly elusive and challenging to combat. It is crucial to promptly develop an effective strategy to address the threats posed by these extensions. A comprehensive review of the research on browser extension vulnerabilities is presented in this paper. The role of malicious links in web browser extensions are examined for several attacks. Detection of malicious browser extension on various aspects are represented namely Intrusion malicious web browser extensions detection using Intrusion detection, Machine learning based detection methods and Deep learning based techniques to mitigate malicious web browser extensions are examined. This study investigates the critical function of malicious detection in protecting web browsers, looking at the changing threats and risk-reduction tactics. A robust cybersecurity frameworks can be created that not only respond to known threats but also anticipate and thwart the strategies of future cyber adversaries by realizing the significance of proactive detection. Thus this survey provides a detailed comparison of various solutions for malicious browser extension.
{"title":"A Holistic Review on Detection of Malicious Browser Extensions and Links using Deep Learning","authors":"Rama Abirami K, Tiago Zonta, Mithileysh Sathiyanarayanan","doi":"10.1109/ICAIC60265.2024.10433842","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433842","url":null,"abstract":"The growth of the Internet has aroused people’s attention toward network security. A secure network environment is fundamental for the expeditious and impeccable development of the Internet. The majority of internet-based tasks can be completed with the help of a web browser. Although many web applications add browser extensions to improve their functionality, some of these extensions are malicious and can access sensitive data without the user’s knowledge. Browser extensions with malicious intent present a growing security concern and have quickly become one of the most prevalent methods used to compromise Internet security. This is largely due to their widespread usage and the extensive privileges they possess. After being installed, these malicious extensions are executed and make an attempt to compromise the victim’s browser. This makes them particularly elusive and challenging to combat. It is crucial to promptly develop an effective strategy to address the threats posed by these extensions. A comprehensive review of the research on browser extension vulnerabilities is presented in this paper. The role of malicious links in web browser extensions are examined for several attacks. Detection of malicious browser extension on various aspects are represented namely Intrusion malicious web browser extensions detection using Intrusion detection, Machine learning based detection methods and Deep learning based techniques to mitigate malicious web browser extensions are examined. This study investigates the critical function of malicious detection in protecting web browsers, looking at the changing threats and risk-reduction tactics. A robust cybersecurity frameworks can be created that not only respond to known threats but also anticipate and thwart the strategies of future cyber adversaries by realizing the significance of proactive detection. Thus this survey provides a detailed comparison of various solutions for malicious browser extension.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"273 4","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139896075","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-07DOI: 10.1109/icaic60265.2024.10433833
{"title":"ICAIC 2024 Cover Page","authors":"","doi":"10.1109/icaic60265.2024.10433833","DOIUrl":"https://doi.org/10.1109/icaic60265.2024.10433833","url":null,"abstract":"","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"34 12","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139965081","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-07DOI: 10.1109/ICAIC60265.2024.10443675
Priyan Malarvizhi Kumar, Kavya Vedantham, Jeeva Selvaraj, B. P. Kavin
The detection of complex attacks by Network Intrusion Detection Systems (NIDS) is hindered by evasion strategies including encrypted traffic and polymorphic malware. Attackers frequently take advantage of holes in NIDS algorithms, emphasising the never-ending cat-and-mouse game between cybersecurity defences and dynamic attack tactics. In the context of cybersecurity, this study offers a sophisticated method for supporting Network Intrusion Detection Systems (NIDS). The tactic includes a thorough preprocessing stage that include functions for normalisation and standardisation in order to recover the accuracy and consistency of the input data. The Perceptive Craving Game Search Optimisation (PCGSO) algorithm is then used for feature selection, maximising the effectiveness of the NIDS. Bidirectional Gated Recurrent Unit (BI-GRU) representations are used in the classification phase because of their ability to identify sequential dependencies in network traffic data. A second PCGSO programme is used to carry out hyperparameter tuning, which guarantees the best possible model performance. The ISCXIDS2012, a popular benchmark dataset in the field, has been selected as the dataset for evaluation. The suggested approach demonstrates how PCGSO may be used to improve feature selection and hyperparameter tweaking, leading to an NIDS that is more accurate and resilient to cyberattacks. Performance evaluations and experimental findings show that the suggested technique outperforms other current models with 99% accuracy
{"title":"Enhanced Network Intrusion Detection System Using PCGSO-Optimized BI-GRU Model in AI-Driven Cybersecurity","authors":"Priyan Malarvizhi Kumar, Kavya Vedantham, Jeeva Selvaraj, B. P. Kavin","doi":"10.1109/ICAIC60265.2024.10443675","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10443675","url":null,"abstract":"The detection of complex attacks by Network Intrusion Detection Systems (NIDS) is hindered by evasion strategies including encrypted traffic and polymorphic malware. Attackers frequently take advantage of holes in NIDS algorithms, emphasising the never-ending cat-and-mouse game between cybersecurity defences and dynamic attack tactics. In the context of cybersecurity, this study offers a sophisticated method for supporting Network Intrusion Detection Systems (NIDS). The tactic includes a thorough preprocessing stage that include functions for normalisation and standardisation in order to recover the accuracy and consistency of the input data. The Perceptive Craving Game Search Optimisation (PCGSO) algorithm is then used for feature selection, maximising the effectiveness of the NIDS. Bidirectional Gated Recurrent Unit (BI-GRU) representations are used in the classification phase because of their ability to identify sequential dependencies in network traffic data. A second PCGSO programme is used to carry out hyperparameter tuning, which guarantees the best possible model performance. The ISCXIDS2012, a popular benchmark dataset in the field, has been selected as the dataset for evaluation. The suggested approach demonstrates how PCGSO may be used to improve feature selection and hyperparameter tweaking, leading to an NIDS that is more accurate and resilient to cyberattacks. Performance evaluations and experimental findings show that the suggested technique outperforms other current models with 99% accuracy","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"34 4","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139965082","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-07DOI: 10.1109/ICAIC60265.2024.10433835
Wenhui Chu, Aobo Jin, Hardik A. Gohel
Designing a robotic system that functions effectively within the specific environment of a Magnetic Resonance Imaging (MRI) scanner requires solving numerous technical issues, such as maintaining the robot’s precision and stability under strong magnetic fields. This research focuses on enhancing MRI’s role in medical imaging, especially in its application to guide intravascular interventions using robot-assisted devices. A newly developed computational system is introduced, designed for seamless integration with the MRI scanner, including a computational unit and user interface. This system processes MR images to delineate the vascular network, establishing virtual paths and boundaries within vessels to prevent procedural damage. Key findings reveal the system’s capability to create tailored magnetic field gradient patterns for device control, considering the vessel’s geometry and safety norms, and adapting to different blood flow characteristics for finer navigation. Additionally, the system’s modeling aspect assesses the safety and feasibility of navigating pre-set vascular paths. Conclusively, this system, based on the Qt framework and C/C++, with specialized software modules, represents a major step forward in merging imaging technology with robotic aid, significantly enhancing precision and safety in intravascular procedures.
{"title":"Simulations and Advancements in MRI-Guided Power-Driven Ferric Tools for Wireless Therapeutic Interventions","authors":"Wenhui Chu, Aobo Jin, Hardik A. Gohel","doi":"10.1109/ICAIC60265.2024.10433835","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433835","url":null,"abstract":"Designing a robotic system that functions effectively within the specific environment of a Magnetic Resonance Imaging (MRI) scanner requires solving numerous technical issues, such as maintaining the robot’s precision and stability under strong magnetic fields. This research focuses on enhancing MRI’s role in medical imaging, especially in its application to guide intravascular interventions using robot-assisted devices. A newly developed computational system is introduced, designed for seamless integration with the MRI scanner, including a computational unit and user interface. This system processes MR images to delineate the vascular network, establishing virtual paths and boundaries within vessels to prevent procedural damage. Key findings reveal the system’s capability to create tailored magnetic field gradient patterns for device control, considering the vessel’s geometry and safety norms, and adapting to different blood flow characteristics for finer navigation. Additionally, the system’s modeling aspect assesses the safety and feasibility of navigating pre-set vascular paths. Conclusively, this system, based on the Qt framework and C/C++, with specialized software modules, represents a major step forward in merging imaging technology with robotic aid, significantly enhancing precision and safety in intravascular procedures.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"13 3","pages":"1-10"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139895374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Conventional processes for analyzing datasets and extracting meaningful information are often time-consuming and laborious. Previous work has identified manual, repetitive coding and data collection as major obstacles that hinder data scientists from undertaking more nuanced labor and high-level projects. To combat this, we evaluated OpenAI’s GPT-3.5 as a "Language Data Scientist" (LDS) that can extrapolate key findings, including correlations and basic information, from a given dataset. The model was tested on a diverse set of benchmark datasets to evaluate its performance across multiple standards, including data science code-generation based tasks involving libraries such as NumPy, Pandas, Scikit-Learn, and TensorFlow, and was broadly successful in correctly answering a given data science query related to the benchmark dataset. The LDS used various novel prompt engineering techniques to effectively answer a given question, including Chain-of-Thought reinforcement and SayCan prompt engineering. Our findings demonstrate great potential for leveraging Large Language Models for low-level, zero-shot data analysis.
{"title":"DataAgent: Evaluating Large Language Models’ Ability to Answer Zero-Shot, Natural Language Queries","authors":"Manit Mishra, Abderrahman Braham, Charles Marsom, Bryan Chung, Gavin Griffin, Dakshesh Sidnerlikar, Chatanya Sarin, Arjun Rajaram","doi":"10.1109/ICAIC60265.2024.10433803","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433803","url":null,"abstract":"Conventional processes for analyzing datasets and extracting meaningful information are often time-consuming and laborious. Previous work has identified manual, repetitive coding and data collection as major obstacles that hinder data scientists from undertaking more nuanced labor and high-level projects. To combat this, we evaluated OpenAI’s GPT-3.5 as a \"Language Data Scientist\" (LDS) that can extrapolate key findings, including correlations and basic information, from a given dataset. The model was tested on a diverse set of benchmark datasets to evaluate its performance across multiple standards, including data science code-generation based tasks involving libraries such as NumPy, Pandas, Scikit-Learn, and TensorFlow, and was broadly successful in correctly answering a given data science query related to the benchmark dataset. The LDS used various novel prompt engineering techniques to effectively answer a given question, including Chain-of-Thought reinforcement and SayCan prompt engineering. Our findings demonstrate great potential for leveraging Large Language Models for low-level, zero-shot data analysis.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"144 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139895527","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-07DOI: 10.1109/ICAIC60265.2024.10433801
Mandar Khoje
In the rapidly evolving landscape of data analytics, safeguarding conversational data privacy presents a pivotal challenge, especially with third-party enterprises commonly offering analytic services. This paper delves into the innovative application of Large Language Models (LLMs) for real-time masking of sensitive information in conversational data. The focus is on balancing privacy protection and data utility for analytics within a multi-stakeholder framework. The significance of data privacy is underscored across sectors, with specific attention to challenges in industries like healthcare, particularly when analytics involve external entities. A comprehensive literature review reveals limitations in existing data masking techniques and explores the role of LLMs in diverse contexts, extending beyond direct healthcare applications.The proposed methodology utilizes LLMs for real-time entity recognition and replacement, effectively masking sensitive information while adhering to privacy regulations. This approach is particularly pertinent for third-party analytics providers dealing with conversational data from various sources. Hypothetical case studies, including healthcare scenarios, showcase the practical application and efficacy of the method in real-world settings with external data analytics providers. The dual assessment evaluates the method’s efficiency in preserving privacy and maintaining data utility for analytical purposes. Experimental results using synthetically generated healthcare conversational data sets further illustrate the effectiveness of the approach in typical third-party analytics service scenarios.The discussion highlights broader implications, addressing challenges and limitations [1] across industries, and emphasizes ethical considerations in handling sensitive data by external entities. In conclusion, the paper summarizes the significant strides achievable with LLMs for data masking, with implications for diverse sectors and analytics providers. Future research directions, especially fine-tuning LLMs for enhanced performance in varied analytic scenarios, are suggested. This study sets the stage for a harmonious coexistence of customer data protection and utility in the intricate ecosystem of data analytics services, facilitated by the advanced capabilities of LLM technology.
{"title":"Navigating Data Privacy and Analytics: The Role of Large Language Models in Masking conversational data in data platforms","authors":"Mandar Khoje","doi":"10.1109/ICAIC60265.2024.10433801","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433801","url":null,"abstract":"In the rapidly evolving landscape of data analytics, safeguarding conversational data privacy presents a pivotal challenge, especially with third-party enterprises commonly offering analytic services. This paper delves into the innovative application of Large Language Models (LLMs) for real-time masking of sensitive information in conversational data. The focus is on balancing privacy protection and data utility for analytics within a multi-stakeholder framework. The significance of data privacy is underscored across sectors, with specific attention to challenges in industries like healthcare, particularly when analytics involve external entities. A comprehensive literature review reveals limitations in existing data masking techniques and explores the role of LLMs in diverse contexts, extending beyond direct healthcare applications.The proposed methodology utilizes LLMs for real-time entity recognition and replacement, effectively masking sensitive information while adhering to privacy regulations. This approach is particularly pertinent for third-party analytics providers dealing with conversational data from various sources. Hypothetical case studies, including healthcare scenarios, showcase the practical application and efficacy of the method in real-world settings with external data analytics providers. The dual assessment evaluates the method’s efficiency in preserving privacy and maintaining data utility for analytical purposes. Experimental results using synthetically generated healthcare conversational data sets further illustrate the effectiveness of the approach in typical third-party analytics service scenarios.The discussion highlights broader implications, addressing challenges and limitations [1] across industries, and emphasizes ethical considerations in handling sensitive data by external entities. In conclusion, the paper summarizes the significant strides achievable with LLMs for data masking, with implications for diverse sectors and analytics providers. Future research directions, especially fine-tuning LLMs for enhanced performance in varied analytic scenarios, are suggested. This study sets the stage for a harmonious coexistence of customer data protection and utility in the intricate ecosystem of data analytics services, facilitated by the advanced capabilities of LLM technology.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"64 2","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139895542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this study, we set out to measure race and gender bias prevalent in text-to-image (TTI) AI image generation, focusing on the popular model Stable Diffusion from Stability AI. Previous investigations into the biases of word embedding models—which serve as the basis for image generation models—have demonstrated that models tend to overstate the relationship between semantic values and gender, ethnicity, or race. These biases are not limited to straightforward stereotypes; more deeply rooted biases may manifest as microaggressions or imposed opinions on policies, such as paid paternity leave decisions. In this analysis, we use image captioning software OpenFlamingo and Stable Diffusion to identify and classify bias within text-to-image models. Utilizing data from the Bureau of Labor Statistics, we engineered 50 prompts for profession and 50 prompts for actions in the interest of coaxing out shallow to systemic biases in the model. Prompts included generating images for "CEO", "nurse", "secretary", "playing basketball", and "doing homework". After generating 20 images for each prompt, we document the model’s results. We find that biases do exist within the model across a variety of prompts. For example, 95% of the images generated for "playing basketball" were African American men. We then analyze our results through categorizing our prompts into a series of income and education levels corresponding to data from the Bureau of Labor Statistics. Ultimately, we find that racial and gender biases are present yet not drastic.
{"title":"Identifying Race and Gender Bias in Stable Diffusion AI Image Generation","authors":"Aadi Chauhan, Taran Anand, Tanisha Jauhari, Arjav Shah, Rudransh Singh, Arjun Rajaram, Rithvik Vanga","doi":"10.1109/ICAIC60265.2024.10433840","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433840","url":null,"abstract":"In this study, we set out to measure race and gender bias prevalent in text-to-image (TTI) AI image generation, focusing on the popular model Stable Diffusion from Stability AI. Previous investigations into the biases of word embedding models—which serve as the basis for image generation models—have demonstrated that models tend to overstate the relationship between semantic values and gender, ethnicity, or race. These biases are not limited to straightforward stereotypes; more deeply rooted biases may manifest as microaggressions or imposed opinions on policies, such as paid paternity leave decisions. In this analysis, we use image captioning software OpenFlamingo and Stable Diffusion to identify and classify bias within text-to-image models. Utilizing data from the Bureau of Labor Statistics, we engineered 50 prompts for profession and 50 prompts for actions in the interest of coaxing out shallow to systemic biases in the model. Prompts included generating images for \"CEO\", \"nurse\", \"secretary\", \"playing basketball\", and \"doing homework\". After generating 20 images for each prompt, we document the model’s results. We find that biases do exist within the model across a variety of prompts. For example, 95% of the images generated for \"playing basketball\" were African American men. We then analyze our results through categorizing our prompts into a series of income and education levels corresponding to data from the Bureau of Labor Statistics. Ultimately, we find that racial and gender biases are present yet not drastic.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"1 3","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139895889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-02-07DOI: 10.1109/ICAIC60265.2024.10433830
Wenhui Chu, Aobo Jin, Hardik A. Gohel
This research aims to develop a novel deep learning network, GBU-Net, utilizing a group-batch-normalized U-Net framework, specifically designed for the precise semantic segmentation of the left ventricle in short-axis cine MRI scans. The methodology includes a down-sampling pathway for feature extraction and an up-sampling pathway for detail restoration, enhanced for medical imaging. Key modifications include techniques for better contextual understanding crucial in cardiac MRI segmentation. The dataset consists of 805 left ventricular MRI scans from 45 patients, with comparative analysis using established metrics such as the dice coefficient and mean perpendicular distance. GBU-Net significantly improves the accuracy of left ventricle segmentation in cine MRI scans. Its innovative design outperforms existing methods in tests, surpassing standard metrics like the dice coefficient and mean perpendicular distance. The approach is unique in its ability to capture contextual information, often missed in traditional CNN-based segmentation. An ensemble of the GBU-Net attains a 97% dice score on the SunnyBrook testing dataset. GBU-Net offers enhanced precision and contextual understanding in left ventricle segmentation for surgical robotics and medical analysis.
{"title":"A Novel Deep Learning Method for Segmenting the Left Ventricle in Cardiac Cine MRI","authors":"Wenhui Chu, Aobo Jin, Hardik A. Gohel","doi":"10.1109/ICAIC60265.2024.10433830","DOIUrl":"https://doi.org/10.1109/ICAIC60265.2024.10433830","url":null,"abstract":"This research aims to develop a novel deep learning network, GBU-Net, utilizing a group-batch-normalized U-Net framework, specifically designed for the precise semantic segmentation of the left ventricle in short-axis cine MRI scans. The methodology includes a down-sampling pathway for feature extraction and an up-sampling pathway for detail restoration, enhanced for medical imaging. Key modifications include techniques for better contextual understanding crucial in cardiac MRI segmentation. The dataset consists of 805 left ventricular MRI scans from 45 patients, with comparative analysis using established metrics such as the dice coefficient and mean perpendicular distance. GBU-Net significantly improves the accuracy of left ventricle segmentation in cine MRI scans. Its innovative design outperforms existing methods in tests, surpassing standard metrics like the dice coefficient and mean perpendicular distance. The approach is unique in its ability to capture contextual information, often missed in traditional CNN-based segmentation. An ensemble of the GBU-Net attains a 97% dice score on the SunnyBrook testing dataset. GBU-Net offers enhanced precision and contextual understanding in left ventricle segmentation for surgical robotics and medical analysis.","PeriodicalId":517265,"journal":{"name":"2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC)","volume":"1 5-6","pages":"1-9"},"PeriodicalIF":0.0,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139896056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：