Heterogeneous systems with analog CMOS circuits integrated with nanoscale memristive devices enable efficient deployment of neural networks on neuromorphic hardware. CMOS Neuron with low footprint can emulate slow temporal dynamics by operating with extremely low current levels. Nevertheless, the current read from the memristive synapses can be higher by several orders of magnitude, and performing impedance matching between neurons and synapses is mandatory. In this paper, we implement an analog leaky integrate and fire (LIF) neuron with a voltage regulator and current attenuator for interfacing CMOS neurons with memristive synapses. In addition, the neuron design proposes a dual leakage that could enable the implementation of local learning rules such as voltage-dependent synaptic plasticity. We also propose a connection scheme to implement adaptive LIF neurons based on two-neuron interaction. The proposed circuits can be used to interface with a variety of synaptic devices and process signals of diverse temporal dynamics.
{"title":"Versatile CMOS Analog LIF Neuron for Memristor-Integrated Neuromorphic Circuits","authors":"Nikhil Garg, Davide Florini, Patrick Dufour, Eloir Muhr, Mathieu Faye, Marc Bocquet, Damien Querlioz, Yann Beilliard, Dominique Drouin, Fabien Alibart, Jean-Michel Portal","doi":"arxiv-2406.19667","DOIUrl":"https://doi.org/arxiv-2406.19667","url":null,"abstract":"Heterogeneous systems with analog CMOS circuits integrated with nanoscale\u0000memristive devices enable efficient deployment of neural networks on\u0000neuromorphic hardware. CMOS Neuron with low footprint can emulate slow temporal\u0000dynamics by operating with extremely low current levels. Nevertheless, the\u0000current read from the memristive synapses can be higher by several orders of\u0000magnitude, and performing impedance matching between neurons and synapses is\u0000mandatory. In this paper, we implement an analog leaky integrate and fire (LIF)\u0000neuron with a voltage regulator and current attenuator for interfacing CMOS\u0000neurons with memristive synapses. In addition, the neuron design proposes a\u0000dual leakage that could enable the implementation of local learning rules such\u0000as voltage-dependent synaptic plasticity. We also propose a connection scheme\u0000to implement adaptive LIF neurons based on two-neuron interaction. The proposed\u0000circuits can be used to interface with a variety of synaptic devices and\u0000process signals of diverse temporal dynamics.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141522519","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}
An Ising model is defined by a quadratic objective function known as the Hamiltonian, composed of spin variables that can take values of either $-1$ or $+1$. The goal is to assign spin values to these variables in a way that minimizes the value of the Hamiltonian. Ising models are instrumental in tackling many combinatorial optimization problems, leading to significant research in developing solvers for them. Notably, D-Wave Systems has pioneered the creation of quantum annealers, programmable solvers based on quantum mechanics, for these models. This paper introduces unit Ising models, where all non-zero coefficients of linear and quadratic terms are either $-1$ or $+1$. Due to the limited resolution of quantum annealers, unit Ising models are more suitable for quantum annealers to find optimal solutions. We propose a novel design methodology for unit Ising models to simulate logic circuits computing Boolean functions through integer linear programming. By optimizing these Ising models with quantum annealers, we can compute Boolean functions and their inverses. With a fixed unit Ising model for a logic circuit, we can potentially design Application-Specific Unit Quantum Annealers (ASUQAs) for computing the inverse function, which is analogous to Application-Specific Integrated Circuits (ASICs) in digital circuitry. For instance, if we apply this technique to a multiplication circuit, we can design an ASUQA for factorization of two numbers. Our findings suggest a powerful new method for compromising the RSA cryptosystem by leveraging ASUQAs in factorization.
{"title":"Designing Unit Ising Models for Logic Gate Simulation through Integer Linear Programming","authors":"Shunsuke Tsukiyama, Koji Nakano, Xiaotian Li, Yasuaki Ito, Takumi Kato, Yuya Kawamata","doi":"arxiv-2406.18130","DOIUrl":"https://doi.org/arxiv-2406.18130","url":null,"abstract":"An Ising model is defined by a quadratic objective function known as the\u0000Hamiltonian, composed of spin variables that can take values of either $-1$ or\u0000$+1$. The goal is to assign spin values to these variables in a way that\u0000minimizes the value of the Hamiltonian. Ising models are instrumental in\u0000tackling many combinatorial optimization problems, leading to significant\u0000research in developing solvers for them. Notably, D-Wave Systems has pioneered\u0000the creation of quantum annealers, programmable solvers based on quantum\u0000mechanics, for these models. This paper introduces unit Ising models, where all\u0000non-zero coefficients of linear and quadratic terms are either $-1$ or $+1$.\u0000Due to the limited resolution of quantum annealers, unit Ising models are more\u0000suitable for quantum annealers to find optimal solutions. We propose a novel\u0000design methodology for unit Ising models to simulate logic circuits computing\u0000Boolean functions through integer linear programming. By optimizing these Ising\u0000models with quantum annealers, we can compute Boolean functions and their\u0000inverses. With a fixed unit Ising model for a logic circuit, we can potentially\u0000design Application-Specific Unit Quantum Annealers (ASUQAs) for computing the\u0000inverse function, which is analogous to Application-Specific Integrated\u0000Circuits (ASICs) in digital circuitry. For instance, if we apply this technique\u0000to a multiplication circuit, we can design an ASUQA for factorization of two\u0000numbers. Our findings suggest a powerful new method for compromising the RSA\u0000cryptosystem by leveraging ASUQAs in factorization.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"110 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141508408","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}
Mauricio Gomes de Queiroz, Paul Jimenez, Raphael Cardoso, Mateus Vidaletti da Costa, Mohab Abdalla, Ian O'Connor, Alberto Bosio, Fabio Pavanello
Photonic Neural Networks (PNNs) are gaining significant interest in the research community due to their potential for high parallelization, low latency, and energy efficiency. PNNs compute using light, which leads to several differences in implementation when compared to electronics, such as the need to represent input features in the photonic domain before feeding them into the network. In this encoding process, it is common to combine multiple features into a single input to reduce the number of inputs and associated devices, leading to smaller and more energy-efficient PNNs. Although this alters the network's handling of input data, its impact on PNNs remains understudied. This paper addresses this open question, investigating the effect of commonly used encoding strategies that combine features on the performance and learning capabilities of PNNs. Here, using the concept of feature importance, we develop a mathematical framework for analyzing feature combination. Through this framework, we demonstrate that encoding multiple features together in a single input determines their relative importance, thus limiting the network's ability to learn from the data. Given some prior knowledge of the data, however, this can also be leveraged for higher accuracy. By selecting an optimal encoding method, we achieve up to a 12.3% improvement in accuracy of PNNs trained on the Iris dataset compared to other encoding techniques, surpassing the performance of networks where features are not combined. These findings highlight the importance of carefully choosing the encoding to the accuracy and decision-making strategies of PNNs, particularly in size or power constrained applications.
{"title":"The Impact of Feature Representation on the Accuracy of Photonic Neural Networks","authors":"Mauricio Gomes de Queiroz, Paul Jimenez, Raphael Cardoso, Mateus Vidaletti da Costa, Mohab Abdalla, Ian O'Connor, Alberto Bosio, Fabio Pavanello","doi":"arxiv-2406.18757","DOIUrl":"https://doi.org/arxiv-2406.18757","url":null,"abstract":"Photonic Neural Networks (PNNs) are gaining significant interest in the\u0000research community due to their potential for high parallelization, low\u0000latency, and energy efficiency. PNNs compute using light, which leads to\u0000several differences in implementation when compared to electronics, such as the\u0000need to represent input features in the photonic domain before feeding them\u0000into the network. In this encoding process, it is common to combine multiple\u0000features into a single input to reduce the number of inputs and associated\u0000devices, leading to smaller and more energy-efficient PNNs. Although this\u0000alters the network's handling of input data, its impact on PNNs remains\u0000understudied. This paper addresses this open question, investigating the effect\u0000of commonly used encoding strategies that combine features on the performance\u0000and learning capabilities of PNNs. Here, using the concept of feature\u0000importance, we develop a mathematical framework for analyzing feature\u0000combination. Through this framework, we demonstrate that encoding multiple\u0000features together in a single input determines their relative importance, thus\u0000limiting the network's ability to learn from the data. Given some prior\u0000knowledge of the data, however, this can also be leveraged for higher accuracy.\u0000By selecting an optimal encoding method, we achieve up to a 12.3% improvement\u0000in accuracy of PNNs trained on the Iris dataset compared to other encoding\u0000techniques, surpassing the performance of networks where features are not\u0000combined. These findings highlight the importance of carefully choosing the\u0000encoding to the accuracy and decision-making strategies of PNNs, particularly\u0000in size or power constrained applications.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141529635","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}
Modern blockchain, such as Ethereum, supports the deployment and execution of so-called smart contracts, autonomous digital programs with significant value of cryptocurrency. Executing smart contracts requires gas costs paid by users, which define the limits of the contract's execution. Logic vulnerabilities in smart contracts can lead to financial losses, and are often the root cause of high-impact cyberattacks. Our objective is threefold: (i) empirically investigate logic vulnerabilities in real-world smart contracts extracted from code changes on GitHub, (ii) introduce Soley, an automated method for detecting logic vulnerabilities in smart contracts, leveraging Large Language Models (LLMs), and (iii) examine mitigation strategies employed by smart contract developers to address these vulnerabilities in real-world scenarios. We obtained smart contracts and related code changes from GitHub. To address the first and third objectives, we qualitatively investigated available logic vulnerabilities using an open coding method. We identified these vulnerabilities and their mitigation strategies. For the second objective, we extracted various logic vulnerabilities, applied preprocessing techniques, and implemented and trained the proposed Soley model. We evaluated Soley along with the performance of various LLMs and compared the results with the state-of-the-art baseline on the task of logic vulnerability detection. From our analysis, we identified nine novel logic vulnerabilities, extending existing taxonomies with these vulnerabilities. Furthermore, we introduced several mitigation strategies extracted from observed developer modifications in real-world scenarios. Our Soley method outperforms existing methods in automatically identifying logic vulnerabilities. Interestingly, the efficacy of LLMs in this task was evident without requiring extensive feature engineering.
{"title":"Soley: Identification and Automated Detection of Logic Vulnerabilities in Ethereum Smart Contracts Using Large Language Models","authors":"Majd Soud, Waltteri Nuutinen, Grischa Liebel","doi":"arxiv-2406.16244","DOIUrl":"https://doi.org/arxiv-2406.16244","url":null,"abstract":"Modern blockchain, such as Ethereum, supports the deployment and execution of\u0000so-called smart contracts, autonomous digital programs with significant value\u0000of cryptocurrency. Executing smart contracts requires gas costs paid by users,\u0000which define the limits of the contract's execution. Logic vulnerabilities in\u0000smart contracts can lead to financial losses, and are often the root cause of\u0000high-impact cyberattacks. Our objective is threefold: (i) empirically\u0000investigate logic vulnerabilities in real-world smart contracts extracted from\u0000code changes on GitHub, (ii) introduce Soley, an automated method for detecting\u0000logic vulnerabilities in smart contracts, leveraging Large Language Models\u0000(LLMs), and (iii) examine mitigation strategies employed by smart contract\u0000developers to address these vulnerabilities in real-world scenarios. We\u0000obtained smart contracts and related code changes from GitHub. To address the\u0000first and third objectives, we qualitatively investigated available logic\u0000vulnerabilities using an open coding method. We identified these\u0000vulnerabilities and their mitigation strategies. For the second objective, we\u0000extracted various logic vulnerabilities, applied preprocessing techniques, and\u0000implemented and trained the proposed Soley model. We evaluated Soley along with\u0000the performance of various LLMs and compared the results with the\u0000state-of-the-art baseline on the task of logic vulnerability detection. From\u0000our analysis, we identified nine novel logic vulnerabilities, extending\u0000existing taxonomies with these vulnerabilities. Furthermore, we introduced\u0000several mitigation strategies extracted from observed developer modifications\u0000in real-world scenarios. Our Soley method outperforms existing methods in\u0000automatically identifying logic vulnerabilities. Interestingly, the efficacy of\u0000LLMs in this task was evident without requiring extensive feature engineering.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"44 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141508412","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}
An Electronic Health Record (EHR) is an electronic database used by healthcare providers to store patients' medical records which may include diagnoses, treatments, costs, and other personal information. Machine learning (ML) algorithms can be used to extract and analyze patient data to improve patient care. Patient records contain highly sensitive information, such as social security numbers (SSNs) and residential addresses, which introduces a need to apply privacy-preserving techniques for these ML models using federated learning and differential privacy.
电子病历（EHR）是医疗服务提供者用来存储患者医疗记录的电子数据库，其中可能包括诊断、治疗、费用和其他个人信息。机器学习（ML）算法可用于提取和分析患者数据，以改善患者护理。患者记录包含高度敏感的信息，如社会安全号（SSN）和住址，这就需要使用联合学习和差异隐私技术为这些 ML 模型应用隐私保护技术。
{"title":"Privacy Preserving Machine Learning for Electronic Health Records using Federated Learning and Differential Privacy","authors":"Naif A. Ganadily, Han J. Xia","doi":"arxiv-2406.15962","DOIUrl":"https://doi.org/arxiv-2406.15962","url":null,"abstract":"An Electronic Health Record (EHR) is an electronic database used by\u0000healthcare providers to store patients' medical records which may include\u0000diagnoses, treatments, costs, and other personal information. Machine learning\u0000(ML) algorithms can be used to extract and analyze patient data to improve\u0000patient care. Patient records contain highly sensitive information, such as\u0000social security numbers (SSNs) and residential addresses, which introduces a\u0000need to apply privacy-preserving techniques for these ML models using federated\u0000learning and differential privacy.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"43 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141522521","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}
Wali Ullah Khan, Chandan Kumar Sheemar, Zaid Abdullah, Eva Lagunas, Symeon Chatzinotas
The terahertz communications have the potential to revolutionize data transfer with unmatched speed and facilitate the development of new high-bandwidth applications. This paper studies the performance of downlink terahertz system assisted by beyond diagonal intelligent reconfigurable surface (BD-IRS). For enhanced energy efficiency and low cost, a joint precoding and BD-IRS phase shift design satisfying the $1$-bit resolution constraints to maximize the spectral efficiency is presented. The original problem is non-linear, NP-hard, and intricately coupled, and obtaining an optimal solution is challenging. To reduce the complexity, we first transform the optimization problem into two problems and then iteratively solve them to achieve an efficient solution. Numerical results demonstrate that the proposed approach for the BD-IRS assisted terahertz system significantly enhances the spectral efficiency compared to the conventional diagonal IRS assisted system.
{"title":"Beyond Diagonal IRS Assisted Ultra Massive THz Systems: A Low Resolution Approach","authors":"Wali Ullah Khan, Chandan Kumar Sheemar, Zaid Abdullah, Eva Lagunas, Symeon Chatzinotas","doi":"arxiv-2406.15880","DOIUrl":"https://doi.org/arxiv-2406.15880","url":null,"abstract":"The terahertz communications have the potential to revolutionize data\u0000transfer with unmatched speed and facilitate the development of new\u0000high-bandwidth applications. This paper studies the performance of downlink\u0000terahertz system assisted by beyond diagonal intelligent reconfigurable surface\u0000(BD-IRS). For enhanced energy efficiency and low cost, a joint precoding and\u0000BD-IRS phase shift design satisfying the $1$-bit resolution constraints to\u0000maximize the spectral efficiency is presented. The original problem is\u0000non-linear, NP-hard, and intricately coupled, and obtaining an optimal solution\u0000is challenging. To reduce the complexity, we first transform the optimization\u0000problem into two problems and then iteratively solve them to achieve an\u0000efficient solution. Numerical results demonstrate that the proposed approach\u0000for the BD-IRS assisted terahertz system significantly enhances the spectral\u0000efficiency compared to the conventional diagonal IRS assisted system.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"68 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141522520","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}
Krish Didwania, Pratinav Seth, Aditya Kasliwal, Amit Agarwal
Agriculture, vital for global sustenance, necessitates innovative solutions due to a lack of organized domain experts, particularly in developing countries where many farmers are impoverished and cannot afford expert consulting. Initiatives like Farmers Helpline play a crucial role in such countries, yet challenges such as high operational costs persist. Automating query resolution can alleviate the burden on traditional call centers, providing farmers with immediate and contextually relevant information. The integration of Agriculture and Artificial Intelligence (AI) offers a transformative opportunity to empower farmers and bridge information gaps. Language models like transformers, the rising stars of AI, possess remarkable language understanding capabilities, making them ideal for addressing information gaps in agriculture. This work explores and demonstrates the transformative potential of Large Language Models (LLMs) in automating query resolution for agricultural farmers, leveraging their expertise in deciphering natural language and understanding context. Using a subset of a vast dataset of real-world farmer queries collected in India, our study focuses on approximately 4 million queries from the state of Tamil Nadu, spanning various sectors, seasonal crops, and query types.
{"title":"AgriLLM: Harnessing Transformers for Farmer Queries","authors":"Krish Didwania, Pratinav Seth, Aditya Kasliwal, Amit Agarwal","doi":"arxiv-2407.04721","DOIUrl":"https://doi.org/arxiv-2407.04721","url":null,"abstract":"Agriculture, vital for global sustenance, necessitates innovative solutions\u0000due to a lack of organized domain experts, particularly in developing countries\u0000where many farmers are impoverished and cannot afford expert consulting.\u0000Initiatives like Farmers Helpline play a crucial role in such countries, yet\u0000challenges such as high operational costs persist. Automating query resolution\u0000can alleviate the burden on traditional call centers, providing farmers with\u0000immediate and contextually relevant information. The integration of Agriculture\u0000and Artificial Intelligence (AI) offers a transformative opportunity to empower\u0000farmers and bridge information gaps. Language models like transformers, the\u0000rising stars of AI, possess remarkable language understanding capabilities,\u0000making them ideal for addressing information gaps in agriculture. This work\u0000explores and demonstrates the transformative potential of Large Language Models\u0000(LLMs) in automating query resolution for agricultural farmers, leveraging\u0000their expertise in deciphering natural language and understanding context.\u0000Using a subset of a vast dataset of real-world farmer queries collected in\u0000India, our study focuses on approximately 4 million queries from the state of\u0000Tamil Nadu, spanning various sectors, seasonal crops, and query types.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"14 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141571838","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}
Child sexual abuse materials (CSAM) pose a significant threat to the safety and well-being of children worldwide. Detecting and preventing the distribution of such materials is a critical task for law enforcement agencies and technology companies. As content moderation is often manual, developing an automated detection system can help reduce human reviewers' exposure to potentially harmful images and accelerate the process of counteracting. This study presents methods for classifying sexually explicit content, which plays a crucial role in the automated CSAM detection system. Several approaches are explored to solve the task: an end-to-end classifier, a classifier with person detection and a private body parts detector. All proposed methods are tested on the images obtained from the online tool for reporting illicit content. Due to legal constraints, access to the data is limited, and all algorithms are executed remotely on the isolated server. The end-to-end classifier yields the most promising results, with an accuracy of 90.17%, after augmenting the training set with the additional neutral samples and adult pornography. While detection-based methods may not achieve higher accuracy rates and cannot serve as a final classifier on their own, their inclusion in the system can be beneficial. Human body-oriented approaches generate results that are easier to interpret, and obtaining more interpretable results is essential when analyzing models that are trained without direct access to data.
{"title":"Detecting sexually explicit content in the context of the child sexual abuse materials (CSAM): end-to-end classifiers and region-based networks","authors":"Weronika Gutfeter, Joanna Gajewska, Andrzej Pacut","doi":"arxiv-2406.14131","DOIUrl":"https://doi.org/arxiv-2406.14131","url":null,"abstract":"Child sexual abuse materials (CSAM) pose a significant threat to the safety\u0000and well-being of children worldwide. Detecting and preventing the distribution\u0000of such materials is a critical task for law enforcement agencies and\u0000technology companies. As content moderation is often manual, developing an\u0000automated detection system can help reduce human reviewers' exposure to\u0000potentially harmful images and accelerate the process of counteracting. This\u0000study presents methods for classifying sexually explicit content, which plays a\u0000crucial role in the automated CSAM detection system. Several approaches are\u0000explored to solve the task: an end-to-end classifier, a classifier with person\u0000detection and a private body parts detector. All proposed methods are tested on\u0000the images obtained from the online tool for reporting illicit content. Due to\u0000legal constraints, access to the data is limited, and all algorithms are\u0000executed remotely on the isolated server. The end-to-end classifier yields the\u0000most promising results, with an accuracy of 90.17%, after augmenting the\u0000training set with the additional neutral samples and adult pornography. While\u0000detection-based methods may not achieve higher accuracy rates and cannot serve\u0000as a final classifier on their own, their inclusion in the system can be\u0000beneficial. Human body-oriented approaches generate results that are easier to\u0000interpret, and obtaining more interpretable results is essential when analyzing\u0000models that are trained without direct access to data.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141522522","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}
Chris Foulon, Marcela Ovando-Tellez, Lia Talozzi, Maurizio Corbetta, Anna Matsulevits, Michel Thiebaut de Schotten
Understanding complex phenomena often requires analyzing high-dimensional data to uncover emergent properties that arise from multifactorial interactions. Here, we present EMUSES (Emerging-properties Mapping Using Spatial Embedding Statistics), an innovative approach employing Uniform Manifold Approximation and Projection (UMAP) to create high-dimensional embeddings that reveal latent structures within data. EMUSES facilitates the exploration and prediction of emergent properties by statistically analyzing these latent spaces. Using three distinct datasets--a handwritten digits dataset from the National Institute of Standards and Technology (NIST, E. Alpaydin, 1998), the Chicago Face Database (Ma et al., 2015), and brain disconnection data post-stroke (Talozzi et al., 2023)--we demonstrate EMUSES' effectiveness in detecting and interpreting emergent properties. Our method not only predicts outcomes with high accuracy but also provides clear visualizations and statistical insights into the underlying interactions within the data. By bridging the gap between predictive accuracy and interpretability, EMUSES offers researchers a powerful tool to understand the multifactorial origins of complex phenomena.
{"title":"Emerging-properties Mapping Using Spatial Embedding Statistics: EMUSES","authors":"Chris Foulon, Marcela Ovando-Tellez, Lia Talozzi, Maurizio Corbetta, Anna Matsulevits, Michel Thiebaut de Schotten","doi":"arxiv-2406.14309","DOIUrl":"https://doi.org/arxiv-2406.14309","url":null,"abstract":"Understanding complex phenomena often requires analyzing high-dimensional\u0000data to uncover emergent properties that arise from multifactorial\u0000interactions. Here, we present EMUSES (Emerging-properties Mapping Using\u0000Spatial Embedding Statistics), an innovative approach employing Uniform\u0000Manifold Approximation and Projection (UMAP) to create high-dimensional\u0000embeddings that reveal latent structures within data. EMUSES facilitates the\u0000exploration and prediction of emergent properties by statistically analyzing\u0000these latent spaces. Using three distinct datasets--a handwritten digits\u0000dataset from the National Institute of Standards and Technology (NIST, E.\u0000Alpaydin, 1998), the Chicago Face Database (Ma et al., 2015), and brain\u0000disconnection data post-stroke (Talozzi et al., 2023)--we demonstrate EMUSES'\u0000effectiveness in detecting and interpreting emergent properties. Our method not\u0000only predicts outcomes with high accuracy but also provides clear\u0000visualizations and statistical insights into the underlying interactions within\u0000the data. By bridging the gap between predictive accuracy and interpretability,\u0000EMUSES offers researchers a powerful tool to understand the multifactorial\u0000origins of complex phenomena.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"8 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141529689","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}
Quantum computing is a cutting-edge field of information technology that harnesses the principles of quantum mechanics to perform computations. It has major implications for the cyber security industry. Existing cyber protection applications are working well, but there are still challenges and vulnerabilities in computer networks. Sometimes data and privacy are also compromised. These complications lead to research questions asking what kind of cyber protection applications of quantum computing are there and what potential methods or techniques can be used for cyber protection? These questions will reveal how much power quantum computing has and to what extent it can outperform the conventional computing systems. This scoping review was conducted by considering 815 papers. It showed the possibilities that can be achievedif quantum technologies are implemented in cyber environments. This scoping review discusses various domains such as algorithms and applications, bioinformatics, cloud and edge computing, the organization of complex systems, application areas focused on security and threats, and the broader quantum computing ecosystem. In each of these areas, there is significant scope for quantum computing to be implemented and to revolutionize the working environment. Numerous quantum computing applications for cyber protection and a number of techniques to protect our data and privacy were identified. The results are not limited to network security but also include data security. This paper also discusses societal aspects, e.g., the applications of quantum computing in the social sciences. This scoping review discusses how to enhance the efficiency and security of quantum computing in various cyber security domains. Additionally, it encourages the reader to think about what kind of techniques and methods can be deployed to secure the cyber world.
{"title":"Cyber Protection Applications of Quantum Computing: A Review","authors":"Ummar Ahmed, Tuomo Sipola, Jari Hautamäki","doi":"arxiv-2406.13259","DOIUrl":"https://doi.org/arxiv-2406.13259","url":null,"abstract":"Quantum computing is a cutting-edge field of information technology that\u0000harnesses the principles of quantum mechanics to perform computations. It has\u0000major implications for the cyber security industry. Existing cyber protection\u0000applications are working well, but there are still challenges and\u0000vulnerabilities in computer networks. Sometimes data and privacy are also\u0000compromised. These complications lead to research questions asking what kind of\u0000cyber protection applications of quantum computing are there and what potential\u0000methods or techniques can be used for cyber protection? These questions will\u0000reveal how much power quantum computing has and to what extent it can\u0000outperform the conventional computing systems. This scoping review was\u0000conducted by considering 815 papers. It showed the possibilities that can be\u0000achievedif quantum technologies are implemented in cyber environments. This\u0000scoping review discusses various domains such as algorithms and applications,\u0000bioinformatics, cloud and edge computing, the organization of complex systems,\u0000application areas focused on security and threats, and the broader quantum\u0000computing ecosystem. In each of these areas, there is significant scope for\u0000quantum computing to be implemented and to revolutionize the working\u0000environment. Numerous quantum computing applications for cyber protection and a\u0000number of techniques to protect our data and privacy were identified. The\u0000results are not limited to network security but also include data security.\u0000This paper also discusses societal aspects, e.g., the applications of quantum\u0000computing in the social sciences. This scoping review discusses how to enhance\u0000the efficiency and security of quantum computing in various cyber security\u0000domains. Additionally, it encourages the reader to think about what kind of\u0000techniques and methods can be deployed to secure the cyber world.","PeriodicalId":501168,"journal":{"name":"arXiv - CS - Emerging Technologies","volume":"54 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141522639","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：