A XAI-based Framework for Frequency Subband Characterization of Cough Spectrograms in Chronic Respiratory Diseasecs.AI updates on arXiv.orgon August 25, 2025 at 4:00 am arXiv:2508.16237v1 Announce Type: cross
Abstract: This paper presents an explainable artificial intelligence (XAI)-based framework for the spectral analysis of cough sounds associated with chronic respiratory diseases, with a particular focus on Chronic Obstructive Pulmonary Disease (COPD). A Convolutional Neural Network (CNN) is trained on time-frequency representations of cough signals, and occlusion maps are used to identify diagnostically relevant regions within the spectrograms. These highlighted areas are subsequently decomposed into five frequency subbands, enabling targeted spectral feature extraction and analysis. The results reveal that spectral patterns differ across subbands and disease groups, uncovering complementary and compensatory trends across the frequency spectrum. Noteworthy, the approach distinguishes COPD from other respiratory conditions, and chronic from non-chronic patient groups, based on interpretable spectral markers. These findings provide insight into the underlying pathophysiological characteristics of cough acoustics and demonstrate the value of frequency-resolved, XAI-enhanced analysis for biomedical signal interpretation and translational respiratory disease diagnostics.
arXiv:2508.16237v1 Announce Type: cross
Abstract: This paper presents an explainable artificial intelligence (XAI)-based framework for the spectral analysis of cough sounds associated with chronic respiratory diseases, with a particular focus on Chronic Obstructive Pulmonary Disease (COPD). A Convolutional Neural Network (CNN) is trained on time-frequency representations of cough signals, and occlusion maps are used to identify diagnostically relevant regions within the spectrograms. These highlighted areas are subsequently decomposed into five frequency subbands, enabling targeted spectral feature extraction and analysis. The results reveal that spectral patterns differ across subbands and disease groups, uncovering complementary and compensatory trends across the frequency spectrum. Noteworthy, the approach distinguishes COPD from other respiratory conditions, and chronic from non-chronic patient groups, based on interpretable spectral markers. These findings provide insight into the underlying pathophysiological characteristics of cough acoustics and demonstrate the value of frequency-resolved, XAI-enhanced analysis for biomedical signal interpretation and translational respiratory disease diagnostics. Read More
On Task Vectors and Gradientscs.AI updates on arXiv.orgon August 25, 2025 at 4:00 am arXiv:2508.16082v1 Announce Type: cross
Abstract: Task arithmetic has emerged as a simple yet powerful technique for model merging, enabling the combination of multiple finetuned models into one. Despite its empirical success, a clear theoretical explanation of why and when it works is lacking. This paper provides a rigorous theoretical foundation for task arithmetic by establishing a connection between task vectors and gradients of the task losses. We show that under standard gradient descent, a task vector generated from one epoch of finetuning is exactly equivalent to the negative gradient of the loss, scaled by the learning rate. For the practical multi-epoch setting, we prove that this equivalence holds approximately, with a second-order error term that we explicitly bound for feed-forward networks. Our empirical analysis across seven vision benchmarks corroborates our theory, demonstrating that the first-epoch gradient dominates the finetuning trajectory in both norm and direction. A key implication is that merging models finetuned for only a single epoch often yields performance comparable to merging fully converged models. These findings reframe task arithmetic as a form of approximate multitask learning, providing a clear rationale for its effectiveness and highlighting the critical role of early training dynamics in model merging.
arXiv:2508.16082v1 Announce Type: cross
Abstract: Task arithmetic has emerged as a simple yet powerful technique for model merging, enabling the combination of multiple finetuned models into one. Despite its empirical success, a clear theoretical explanation of why and when it works is lacking. This paper provides a rigorous theoretical foundation for task arithmetic by establishing a connection between task vectors and gradients of the task losses. We show that under standard gradient descent, a task vector generated from one epoch of finetuning is exactly equivalent to the negative gradient of the loss, scaled by the learning rate. For the practical multi-epoch setting, we prove that this equivalence holds approximately, with a second-order error term that we explicitly bound for feed-forward networks. Our empirical analysis across seven vision benchmarks corroborates our theory, demonstrating that the first-epoch gradient dominates the finetuning trajectory in both norm and direction. A key implication is that merging models finetuned for only a single epoch often yields performance comparable to merging fully converged models. These findings reframe task arithmetic as a form of approximate multitask learning, providing a clear rationale for its effectiveness and highlighting the critical role of early training dynamics in model merging. Read More
PediatricsMQA: a Multi-modal Pediatrics Question Answering Benchmarkcs.AI updates on arXiv.orgon August 25, 2025 at 4:00 am arXiv:2508.16439v1 Announce Type: cross
Abstract: Large language models (LLMs) and vision-augmented LLMs (VLMs) have significantly advanced medical informatics, diagnostics, and decision support. However, these models exhibit systematic biases, particularly age bias, compromising their reliability and equity. This is evident in their poorer performance on pediatric-focused text and visual question-answering tasks. This bias reflects a broader imbalance in medical research, where pediatric studies receive less funding and representation despite the significant disease burden in children. To address these issues, a new comprehensive multi-modal pediatric question-answering benchmark, PediatricsMQA, has been introduced. It consists of 3,417 text-based multiple-choice questions (MCQs) covering 131 pediatric topics across seven developmental stages (prenatal to adolescent) and 2,067 vision-based MCQs using 634 pediatric images from 67 imaging modalities and 256 anatomical regions. The dataset was developed using a hybrid manual-automatic pipeline, incorporating peer-reviewed pediatric literature, validated question banks, existing benchmarks, and existing QA resources. Evaluating state-of-the-art open models, we find dramatic performance drops in younger cohorts, highlighting the need for age-aware methods to ensure equitable AI support in pediatric care.
arXiv:2508.16439v1 Announce Type: cross
Abstract: Large language models (LLMs) and vision-augmented LLMs (VLMs) have significantly advanced medical informatics, diagnostics, and decision support. However, these models exhibit systematic biases, particularly age bias, compromising their reliability and equity. This is evident in their poorer performance on pediatric-focused text and visual question-answering tasks. This bias reflects a broader imbalance in medical research, where pediatric studies receive less funding and representation despite the significant disease burden in children. To address these issues, a new comprehensive multi-modal pediatric question-answering benchmark, PediatricsMQA, has been introduced. It consists of 3,417 text-based multiple-choice questions (MCQs) covering 131 pediatric topics across seven developmental stages (prenatal to adolescent) and 2,067 vision-based MCQs using 634 pediatric images from 67 imaging modalities and 256 anatomical regions. The dataset was developed using a hybrid manual-automatic pipeline, incorporating peer-reviewed pediatric literature, validated question banks, existing benchmarks, and existing QA resources. Evaluating state-of-the-art open models, we find dramatic performance drops in younger cohorts, highlighting the need for age-aware methods to ensure equitable AI support in pediatric care. Read More
What happens when AI data centres run out of space? NVIDIA’s new solution explainedAI Newson August 25, 2025 at 9:00 am When AI data centres run out of space, they face a costly dilemma: build bigger facilities or find ways to make multiple locations work together seamlessly. NVIDIA’s latest Spectrum-XGS Ethernet technology promises to solve this challenge by connecting AI data centres across vast distances into what the company calls “giga-scale AI super-factories.” Announced ahead of Hot
The post What happens when AI data centres run out of space? NVIDIA’s new solution explained appeared first on AI News.
When AI data centres run out of space, they face a costly dilemma: build bigger facilities or find ways to make multiple locations work together seamlessly. NVIDIA’s latest Spectrum-XGS Ethernet technology promises to solve this challenge by connecting AI data centres across vast distances into what the company calls “giga-scale AI super-factories.” Announced ahead of Hot
The post What happens when AI data centres run out of space? NVIDIA’s new solution explained appeared first on AI News. Read More
The US federal government secures a massive Google Gemini AI deal at $0.47 per agencyAI Newson August 25, 2025 at 8:00 am Google Gemini will soon power federal operations across the United States government following a sweeping new agreement between the General Services Administration (GSA) and Google that delivers comprehensive AI capabilities at unprecedented pricing. The “Gemini for Government” offering, announced by GSA, represents one of the most significant government AI procurement deals to date. Under the OneGov agreement
The post The US federal government secures a massive Google Gemini AI deal at $0.47 per agency appeared first on AI News.
Google Gemini will soon power federal operations across the United States government following a sweeping new agreement between the General Services Administration (GSA) and Google that delivers comprehensive AI capabilities at unprecedented pricing. The “Gemini for Government” offering, announced by GSA, represents one of the most significant government AI procurement deals to date. Under the OneGov agreement
The post The US federal government secures a massive Google Gemini AI deal at $0.47 per agency appeared first on AI News. Read More
Can Large Language Models Simulate Human Responses? A Case Study of Stated Preference Experiments in the Context of Heating-related Choicescs.AI updates on arXiv.orgon August 25, 2025 at 4:00 am arXiv:2503.10652v3 Announce Type: replace-cross
Abstract: Stated preference (SP) surveys are a key method to research how individuals make trade-offs in hypothetical, also futuristic, scenarios. In energy context this includes key decarbonisation enablement contexts, such as low-carbon technologies, distributed renewable energy generation, and demand-side response [1,2]. However, they tend to be costly, time-consuming, and can be affected by respondent fatigue and ethical constraints. Large language models (LLMs) have demonstrated remarkable capabilities in generating human-like textual responses, prompting growing interest in their application to survey research. This study investigates the use of LLMs to simulate consumer choices in energy-related SP surveys and explores their integration into data analysis workflows. A series of test scenarios were designed to systematically assess the simulation performance of several LLMs (LLaMA 3.1, Mistral, GPT-3.5 and DeepSeek-R1) at both individual and aggregated levels, considering contexts factors such as prompt design, in-context learning (ICL), chain-of-thought (CoT) reasoning, LLM types, integration with traditional choice models, and potential biases. Cloud-based LLMs do not consistently outperform smaller local models. In this study, the reasoning model DeepSeek-R1 achieves the highest average accuracy (77%) and outperforms non-reasoning LLMs in accuracy, factor identification, and choice distribution alignment. Across models, systematic biases are observed against the gas boiler and no-retrofit options, with a preference for more energy-efficient alternatives. The findings suggest that previous SP choices are the most effective input factor, while longer prompts with additional factors and varied formats can cause LLMs to lose focus, reducing accuracy.
arXiv:2503.10652v3 Announce Type: replace-cross
Abstract: Stated preference (SP) surveys are a key method to research how individuals make trade-offs in hypothetical, also futuristic, scenarios. In energy context this includes key decarbonisation enablement contexts, such as low-carbon technologies, distributed renewable energy generation, and demand-side response [1,2]. However, they tend to be costly, time-consuming, and can be affected by respondent fatigue and ethical constraints. Large language models (LLMs) have demonstrated remarkable capabilities in generating human-like textual responses, prompting growing interest in their application to survey research. This study investigates the use of LLMs to simulate consumer choices in energy-related SP surveys and explores their integration into data analysis workflows. A series of test scenarios were designed to systematically assess the simulation performance of several LLMs (LLaMA 3.1, Mistral, GPT-3.5 and DeepSeek-R1) at both individual and aggregated levels, considering contexts factors such as prompt design, in-context learning (ICL), chain-of-thought (CoT) reasoning, LLM types, integration with traditional choice models, and potential biases. Cloud-based LLMs do not consistently outperform smaller local models. In this study, the reasoning model DeepSeek-R1 achieves the highest average accuracy (77%) and outperforms non-reasoning LLMs in accuracy, factor identification, and choice distribution alignment. Across models, systematic biases are observed against the gas boiler and no-retrofit options, with a preference for more energy-efficient alternatives. The findings suggest that previous SP choices are the most effective input factor, while longer prompts with additional factors and varied formats can cause LLMs to lose focus, reducing accuracy. Read More
Rachel James, AbbVie: Harnessing AI for corporate cybersecurityAI Newson August 22, 2025 at 2:48 pm Cybersecurity is in the midst of a fresh arms race, and the powerful weapon of choice in this new era is AI. AI offers a classic double-edged sword: a powerful shield for defenders and a potent new tool for those with malicious intent. Navigating this complex battleground requires a steady hand and a deep understanding
The post Rachel James, AbbVie: Harnessing AI for corporate cybersecurity appeared first on AI News.
Cybersecurity is in the midst of a fresh arms race, and the powerful weapon of choice in this new era is AI. AI offers a classic double-edged sword: a powerful shield for defenders and a potent new tool for those with malicious intent. Navigating this complex battleground requires a steady hand and a deep understanding
The post Rachel James, AbbVie: Harnessing AI for corporate cybersecurity appeared first on AI News. Read More
The Download: Google’s AI energy expenditure, and handing over DNA data to the policeMIT Technology Reviewon August 22, 2025 at 12:10 pm This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. In a first, Google has released data on how much energy an AI prompt uses Google has just released a report detailing how much energy its Gemini apps use for each query. In…
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. In a first, Google has released data on how much energy an AI prompt uses Google has just released a report detailing how much energy its Gemini apps use for each query. In… Read More
The case against humans in spaceMIT Technology Reviewon August 22, 2025 at 10:00 am Elon Musk and Jeff Bezos are bitter rivals in the commercial space race, but they agree on one thing: Settling space is an existential imperative. Space is the place. The final frontier. It is our human destiny to transcend our home world and expand our civilization to extraterrestrial vistas. This belief has been mainstream for…
Elon Musk and Jeff Bezos are bitter rivals in the commercial space race, but they agree on one thing: Settling space is an existential imperative. Space is the place. The final frontier. It is our human destiny to transcend our home world and expand our civilization to extraterrestrial vistas. This belief has been mainstream for… Read More
A Systematic Survey of Model Extraction Attacks and Defenses: State-of-the-Art and Perspectivescs.AI updates on arXiv.orgon August 22, 2025 at 4:00 am arXiv:2508.15031v1 Announce Type: cross
Abstract: Machine learning (ML) models have significantly grown in complexity and utility, driving advances across multiple domains. However, substantial computational resources and specialized expertise have historically restricted their wide adoption. Machine-Learning-as-a-Service (MLaaS) platforms have addressed these barriers by providing scalable, convenient, and affordable access to sophisticated ML models through user-friendly APIs. While this accessibility promotes widespread use of advanced ML capabilities, it also introduces vulnerabilities exploited through Model Extraction Attacks (MEAs). Recent studies have demonstrated that adversaries can systematically replicate a target model’s functionality by interacting with publicly exposed interfaces, posing threats to intellectual property, privacy, and system security. In this paper, we offer a comprehensive survey of MEAs and corresponding defense strategies. We propose a novel taxonomy that classifies MEAs according to attack mechanisms, defense approaches, and computing environments. Our analysis covers various attack techniques, evaluates their effectiveness, and highlights challenges faced by existing defenses, particularly the critical trade-off between preserving model utility and ensuring security. We further assess MEAs within different computing paradigms and discuss their technical, ethical, legal, and societal implications, along with promising directions for future research. This systematic survey aims to serve as a valuable reference for researchers, practitioners, and policymakers engaged in AI security and privacy. Additionally, we maintain an online repository continuously updated with related literature at https://github.com/kzhao5/ModelExtractionPapers.
arXiv:2508.15031v1 Announce Type: cross
Abstract: Machine learning (ML) models have significantly grown in complexity and utility, driving advances across multiple domains. However, substantial computational resources and specialized expertise have historically restricted their wide adoption. Machine-Learning-as-a-Service (MLaaS) platforms have addressed these barriers by providing scalable, convenient, and affordable access to sophisticated ML models through user-friendly APIs. While this accessibility promotes widespread use of advanced ML capabilities, it also introduces vulnerabilities exploited through Model Extraction Attacks (MEAs). Recent studies have demonstrated that adversaries can systematically replicate a target model’s functionality by interacting with publicly exposed interfaces, posing threats to intellectual property, privacy, and system security. In this paper, we offer a comprehensive survey of MEAs and corresponding defense strategies. We propose a novel taxonomy that classifies MEAs according to attack mechanisms, defense approaches, and computing environments. Our analysis covers various attack techniques, evaluates their effectiveness, and highlights challenges faced by existing defenses, particularly the critical trade-off between preserving model utility and ensuring security. We further assess MEAs within different computing paradigms and discuss their technical, ethical, legal, and societal implications, along with promising directions for future research. This systematic survey aims to serve as a valuable reference for researchers, practitioners, and policymakers engaged in AI security and privacy. Additionally, we maintain an online repository continuously updated with related literature at https://github.com/kzhao5/ModelExtractionPapers. Read More