Recently, there has been a growing interest in improving students' competitiveness in STEM education. Self-reporting and observation are the most used tools for the assessment of STEM education. Despite their effectiveness, such assessment tools face several challenges, such as being labor-intensive and time-consuming, prone to subjective awareness, depending on memory limitations, and being influenced due to social expectations. To address these challenges, in this research, we propose an approach called Image4Assess that---by benefiting from state-of-the-art machine learning like convolutional neural networks and transfer learning---automatically and uninterruptedly assesses students' learning processes during STEM activities using image processing. Our findings reveal that the Image4Assess approach can achieve accuracy, precision, and recall higher than 85% in the learning process recognition of students. This implies that it is feasible to accurately measure the learning process of students in STEM education using their imagery data. We also found that there is a significant correlation between the learning processes automatically identified by our proposed approach and students' post-test, confirming the effectiveness of the proposed approach in real-world classrooms.
{"title":"Image4Assess: Automatic learning processes recognition using image processing","authors":"Hsin-Yu Lee, Maral Hooshyar, Chia-Ju Lin, Wei-Sheng Wang, Yueh-Min Huang","doi":"10.1145/3555776.3577643","DOIUrl":"https://doi.org/10.1145/3555776.3577643","url":null,"abstract":"Recently, there has been a growing interest in improving students' competitiveness in STEM education. Self-reporting and observation are the most used tools for the assessment of STEM education. Despite their effectiveness, such assessment tools face several challenges, such as being labor-intensive and time-consuming, prone to subjective awareness, depending on memory limitations, and being influenced due to social expectations. To address these challenges, in this research, we propose an approach called Image4Assess that---by benefiting from state-of-the-art machine learning like convolutional neural networks and transfer learning---automatically and uninterruptedly assesses students' learning processes during STEM activities using image processing. Our findings reveal that the Image4Assess approach can achieve accuracy, precision, and recall higher than 85% in the learning process recognition of students. This implies that it is feasible to accurately measure the learning process of students in STEM education using their imagery data. We also found that there is a significant correlation between the learning processes automatically identified by our proposed approach and students' post-test, confirming the effectiveness of the proposed approach in real-world classrooms.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"27 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74765397","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}
HotStuff is a Byzantine fault-tolerant state machine replication protocol that incurs linear communication costs to achieve consensus. This linear scalability promoted the protocol to be adopted as the consensus mechanism in permissioned blockchains. This paper discusses the architecture and evaluation of our extensible framework to implement three HotStuff variants. This reimplementation demonstrates the extensibility of our framework to implement other HotStuff-like protocols. Leveraging our deployment tool, we evaluated our implementation on a wide variety of configurations.
{"title":"An Extensible Framework for Implementing Byzantine Fault-Tolerant Protocols","authors":"Hanish Gogada, J. Olsen, H. Meling, Leander Jehl","doi":"10.1145/3555776.3578614","DOIUrl":"https://doi.org/10.1145/3555776.3578614","url":null,"abstract":"HotStuff is a Byzantine fault-tolerant state machine replication protocol that incurs linear communication costs to achieve consensus. This linear scalability promoted the protocol to be adopted as the consensus mechanism in permissioned blockchains. This paper discusses the architecture and evaluation of our extensible framework to implement three HotStuff variants. This reimplementation demonstrates the extensibility of our framework to implement other HotStuff-like protocols. Leveraging our deployment tool, we evaluated our implementation on a wide variety of configurations.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"17 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83409586","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}
Prabir Mondal, Daipayan Chakder, Subham Raj, S. Saha, N. Onoe
The Recommendation System (RS) development and recommending customers' preferred products to the customer are highly desirable motives in today's digital market. Most of the RSs are mainly based on textual information of the engaged entities in the platform and the ratings provided by the users to the products. This paper develops a movie recommendation system where the cold-start problem relating to rating information dependency has been dealt with and the multi-modality approach is introduced. The proposed method differs from existing approaches in three main aspects: (a) implementation of knowledge graph for text embedding, (b) besides textual information, other modalities of movies like video, and audio are employed rather than rating information for generating movie/user representation and this approach deals with the cold-start problem effectively, (c) utilization of graph convolutional network (GCN) for generating some further hidden features and also for developing regression system.
{"title":"Graph Convolutional Neural Network for Multimodal Movie Recommendation","authors":"Prabir Mondal, Daipayan Chakder, Subham Raj, S. Saha, N. Onoe","doi":"10.1145/3555776.3577853","DOIUrl":"https://doi.org/10.1145/3555776.3577853","url":null,"abstract":"The Recommendation System (RS) development and recommending customers' preferred products to the customer are highly desirable motives in today's digital market. Most of the RSs are mainly based on textual information of the engaged entities in the platform and the ratings provided by the users to the products. This paper develops a movie recommendation system where the cold-start problem relating to rating information dependency has been dealt with and the multi-modality approach is introduced. The proposed method differs from existing approaches in three main aspects: (a) implementation of knowledge graph for text embedding, (b) besides textual information, other modalities of movies like video, and audio are employed rather than rating information for generating movie/user representation and this approach deals with the cold-start problem effectively, (c) utilization of graph convolutional network (GCN) for generating some further hidden features and also for developing regression system.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"96 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76865314","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 key challenges that recommendation systems must overcome are data isolation and privacy protection issues. Federated learning can efficiently train global models using decentralized data while preserving privacy. In real-world applications, however, it is difficult to achieve high prediction accuracy due to the heterogeneity of devices, the lack of data, and the limited generalization capacity of models. In this research, we introduce a personalized federated knowledge distillation model for a recommendation system based on a multi-head attention mechanism for recommendation systems. Specifically, we first employ federated distillation to improve the performance of student models and introduce a multi-head attention mechanism to enhance user encoding information. Next, we incorporate Wasserstein distance into the objective function of combined distillation to reduce the distribution gap between teacher and student networks and also use an adaptive learning rate technique to enhance convergence. We show that the proposed approach achieves better effectiveness and robustness through benchmarks.
{"title":"MAFD: A Federated Distillation Approach with Multi-head Attention for Recommendation Tasks","authors":"Aming Wu, Young-Woo Kwon","doi":"10.1145/3555776.3577849","DOIUrl":"https://doi.org/10.1145/3555776.3577849","url":null,"abstract":"The key challenges that recommendation systems must overcome are data isolation and privacy protection issues. Federated learning can efficiently train global models using decentralized data while preserving privacy. In real-world applications, however, it is difficult to achieve high prediction accuracy due to the heterogeneity of devices, the lack of data, and the limited generalization capacity of models. In this research, we introduce a personalized federated knowledge distillation model for a recommendation system based on a multi-head attention mechanism for recommendation systems. Specifically, we first employ federated distillation to improve the performance of student models and introduce a multi-head attention mechanism to enhance user encoding information. Next, we incorporate Wasserstein distance into the objective function of combined distillation to reduce the distribution gap between teacher and student networks and also use an adaptive learning rate technique to enhance convergence. We show that the proposed approach achieves better effectiveness and robustness through benchmarks.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"13 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76952591","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}
Embedded Systems (ES) are present in several domains like automotive, smart homes, smart cities, industry, and healthcare, to name but a few. ES brings new challenges to designing embedded software that requires a high level of abstraction and being aware of resource consumption, mainly on resource-constrained devices. Modern programming languages like JavaScript (JS) can help solve these issues. However, JS is an interpreted language that demands attention to develop applications considering the balance between performance and resource consumption. In this scenario, this paper introduces an architecture design that proposes to model software for embedded systems as event-driven applications. Our design combines traditional architectures traits of Time-triggered (TT) and Event-triggered (ET) into a framework named JSEVAsync, promoting a hybrid system that explores JavaScript's non-blocking concept as a development interface to structure the algorithms into asynchronous units. As a result, we aid the development of applications with high abstraction levels and better resource consumption. To validate it, we compare C- and JavaScript-based applications, analyze the source code (static code analysis) to extract software quality metrics, and explore the results from the energy consumption perspective. We found that writing code through JSEVAsync can be up to 21% more energy efficient than the traditional method and can improve design-time metrics.
{"title":"Student Research Abstract: A Hybrid Approach to Design Embedded Software Using JavaScript's Non-blocking Principle","authors":"Fernando L. Oliveira","doi":"10.1145/3555776.3577210","DOIUrl":"https://doi.org/10.1145/3555776.3577210","url":null,"abstract":"Embedded Systems (ES) are present in several domains like automotive, smart homes, smart cities, industry, and healthcare, to name but a few. ES brings new challenges to designing embedded software that requires a high level of abstraction and being aware of resource consumption, mainly on resource-constrained devices. Modern programming languages like JavaScript (JS) can help solve these issues. However, JS is an interpreted language that demands attention to develop applications considering the balance between performance and resource consumption. In this scenario, this paper introduces an architecture design that proposes to model software for embedded systems as event-driven applications. Our design combines traditional architectures traits of Time-triggered (TT) and Event-triggered (ET) into a framework named JSEVAsync, promoting a hybrid system that explores JavaScript's non-blocking concept as a development interface to structure the algorithms into asynchronous units. As a result, we aid the development of applications with high abstraction levels and better resource consumption. To validate it, we compare C- and JavaScript-based applications, analyze the source code (static code analysis) to extract software quality metrics, and explore the results from the energy consumption perspective. We found that writing code through JSEVAsync can be up to 21% more energy efficient than the traditional method and can improve design-time metrics.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"44 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79933869","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}
Shakthi Weerasinghe, A. Zaslavsky, S. Loke, A. Abken, A. Hassani, A. Medvedev
We contend that performance metrics-driven adaptive context caching has a profound impact on performance efficiency in distributed context management systems (CMS). This paper proposes an adaptive context caching approach based on (i) a model of economics-inspired expected returns of caching particular items, and (ii) learning from historical context caching performance, i.e., our approach adaptively (with respect to statistics on historical performance) caches "context" with the objective of minimizing the cost incurred by a CMS in responding to context queries. Our novel algorithm enables context queries and sub-queries to reuse and repurpose cached context in an efficient manner, different from traditional data caching. The paper also proposes heuristics and adaptive policies such as eviction and context cache memory scaling. The method is evaluated using a synthetically generated load of sub-queries inspired by a real-world scenario. We further investigate optimal adaptive caching configurations under different settings. This paper presents and discusses our findings that the proposed statistical selective caching method reaches short-term cost optimality fast under massively volatile queries. The proposed method outperforms related algorithms by up to 47.9% in cost efficiency.
{"title":"Adaptive Context Caching for Efficient Distributed Context Management Systems","authors":"Shakthi Weerasinghe, A. Zaslavsky, S. Loke, A. Abken, A. Hassani, A. Medvedev","doi":"10.1145/3555776.3577602","DOIUrl":"https://doi.org/10.1145/3555776.3577602","url":null,"abstract":"We contend that performance metrics-driven adaptive context caching has a profound impact on performance efficiency in distributed context management systems (CMS). This paper proposes an adaptive context caching approach based on (i) a model of economics-inspired expected returns of caching particular items, and (ii) learning from historical context caching performance, i.e., our approach adaptively (with respect to statistics on historical performance) caches \"context\" with the objective of minimizing the cost incurred by a CMS in responding to context queries. Our novel algorithm enables context queries and sub-queries to reuse and repurpose cached context in an efficient manner, different from traditional data caching. The paper also proposes heuristics and adaptive policies such as eviction and context cache memory scaling. The method is evaluated using a synthetically generated load of sub-queries inspired by a real-world scenario. We further investigate optimal adaptive caching configurations under different settings. This paper presents and discusses our findings that the proposed statistical selective caching method reaches short-term cost optimality fast under massively volatile queries. The proposed method outperforms related algorithms by up to 47.9% in cost efficiency.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"22 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90098562","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}
Stock-index movement prediction is an important research topic in FinTech because the index indicates the economic status of a whole country. With a set of daily candlesticks of the stock-index, investors could gain a meaningful basis for the prediction of the next day's movement. This paper proposes a stock-index price-movement prediction model, Combined Time-View TabNet (CTV-TabNet), a novel approach that utilizes attributes of the candlesticks data with multi-time windows. Our model comprises three modules: TabNet encoder, gated recurrent unit with a sequence control, and multi-time combiner. They work together to forecast the movements based on the sequential attributes of the candlesticks. CTV-TabNet not only outperforms baseline models in prediction performance on 20 stock-indices of 14 different countries but also yields higher returns of index-futures trading simulations when compared to the baselines. Additionally, our model provides comprehensive interpretations of the stock-index related to its inherent properties in predictive performance.
{"title":"Exploring Candlesticks and Multi-Time Windows for Forecasting Stock-Index Movements","authors":"Kanghyeon Seo, Jihoon Yang","doi":"10.1145/3555776.3577604","DOIUrl":"https://doi.org/10.1145/3555776.3577604","url":null,"abstract":"Stock-index movement prediction is an important research topic in FinTech because the index indicates the economic status of a whole country. With a set of daily candlesticks of the stock-index, investors could gain a meaningful basis for the prediction of the next day's movement. This paper proposes a stock-index price-movement prediction model, Combined Time-View TabNet (CTV-TabNet), a novel approach that utilizes attributes of the candlesticks data with multi-time windows. Our model comprises three modules: TabNet encoder, gated recurrent unit with a sequence control, and multi-time combiner. They work together to forecast the movements based on the sequential attributes of the candlesticks. CTV-TabNet not only outperforms baseline models in prediction performance on 20 stock-indices of 14 different countries but also yields higher returns of index-futures trading simulations when compared to the baselines. Additionally, our model provides comprehensive interpretations of the stock-index related to its inherent properties in predictive performance.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"38 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86773524","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 the days of AI, data-centric machine learning (ML) models are increasingly used in various complex systems. While many researchers are focusing on specifying ML-specific performance requirements, not enough guideline is provided to engineer the data requirements systematically involving diverse stakeholders. Lack of written agreement about the training data, collaboration bottlenecks, lack of data validation framework, etc. are posing new challenges to ensuring training data fitness for safety-critical ML components. To reduce these gaps, we propose a multi-layered framework that helps to perceive and elicit data requirements. We provide a template for verifiable data requirements specifications. Moreover, we show how such requirements can facilitate an evidence-driven assessment of the training data quality based on the experts' judgments about the satisfaction of the requirements. We use Dempster Shafer's theory to combine experts' subjective opinions in the process. A preliminary case study on the CityPersons dataset for the pedestrian detection feature of autonomous cars shows the usefulness of the proposed framework for data requirements understanding and the confidence assessment of the dataset.
{"title":"A Multi-layered Collaborative Framework for Evidence-driven Data Requirements Engineering for Machine Learning-based Safety-critical Systems","authors":"Sangeeta Dey, Seok-Won Lee","doi":"10.1145/3555776.3577647","DOIUrl":"https://doi.org/10.1145/3555776.3577647","url":null,"abstract":"In the days of AI, data-centric machine learning (ML) models are increasingly used in various complex systems. While many researchers are focusing on specifying ML-specific performance requirements, not enough guideline is provided to engineer the data requirements systematically involving diverse stakeholders. Lack of written agreement about the training data, collaboration bottlenecks, lack of data validation framework, etc. are posing new challenges to ensuring training data fitness for safety-critical ML components. To reduce these gaps, we propose a multi-layered framework that helps to perceive and elicit data requirements. We provide a template for verifiable data requirements specifications. Moreover, we show how such requirements can facilitate an evidence-driven assessment of the training data quality based on the experts' judgments about the satisfaction of the requirements. We use Dempster Shafer's theory to combine experts' subjective opinions in the process. A preliminary case study on the CityPersons dataset for the pedestrian detection feature of autonomous cars shows the usefulness of the proposed framework for data requirements understanding and the confidence assessment of the dataset.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"35 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89390161","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 people are used to recording more and more videos using camera applications for keeping and sharing their life on social media and video-sharing platforms. To capture extensive multimedia materials, reducing the power consumption of recorded videos from camera applications plays an important role for user experience of mobile devices. This paper studies how to process and display power-saving videos recorded by camera applications on mobile devices in a real-time manner. Based on pixel-scaling methods, we design an appropriate feature map and adopt a visual attention model under the real-time limitation to effectively access attention distribution. Then, based on segmentation properties, a parallel design is appropriately applied to exploit available computation power. Next, we propose a frame-ratio predictor using machine-learning methods to efficiently predict frame ratios in a frame. Finally, the results of the comprehensive experiments conducted on a commercial smartphone with four real-world videos to evaluate the performance of the proposed design are very encouraging.
{"title":"Exploiting Machine-learning Prediction for Enabling Real-time Pixel-scaling Techniques in Mobile Camera Applications","authors":"S. Wei, Sheng-Da Tsai, Chun-Han Lin","doi":"10.1145/3555776.3577770","DOIUrl":"https://doi.org/10.1145/3555776.3577770","url":null,"abstract":"Modern people are used to recording more and more videos using camera applications for keeping and sharing their life on social media and video-sharing platforms. To capture extensive multimedia materials, reducing the power consumption of recorded videos from camera applications plays an important role for user experience of mobile devices. This paper studies how to process and display power-saving videos recorded by camera applications on mobile devices in a real-time manner. Based on pixel-scaling methods, we design an appropriate feature map and adopt a visual attention model under the real-time limitation to effectively access attention distribution. Then, based on segmentation properties, a parallel design is appropriately applied to exploit available computation power. Next, we propose a frame-ratio predictor using machine-learning methods to efficiently predict frame ratios in a frame. Finally, the results of the comprehensive experiments conducted on a commercial smartphone with four real-world videos to evaluate the performance of the proposed design are very encouraging.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"13 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87027295","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}
Unsupervised side-channel attacks allow extracting secret keys manipulated by cryptographic primitives through leakages of their physical implementations. As opposed to supervised attacks, they do not require a preliminary profiling of the target, constituting a broader threat since they imply weaker assumptions on the adversary model. Their downside is their requirement for some a priori knowledge on the leakage model of the device. On one hand, stochastic attacks such as the Linear Regression Analysis (LRA) allow for a flexible a priori, but are mostly limited to a univariate treatment of the traces. On the other hand, model-based attacks require an explicit formulation of the leakage model but have recently been extended to multidimensional versions allowing to benefit from the potential of Deep Learning (DL) techniques. The EVIL Machine Attack (EMA), introduced in this paper, aims at taking the best of both worlds. Inspired by generative adversarial networks, its architecture is able to recover a representation of the leakage model, which is then turned into a key distinguisher allowing flexible a priori. In addition, state-of-the-art DL techniques require 256 network trainings to conduct the attack. EMA requires only one, scaling down the time complexity of such attacks by a considerable factor. Simulations and real experiments show that EMA is applicable in cases where the adversary has very low knowledge on the leakage model, while significantly reducing the required number of traces compared to a classical LRA. Eventually, a generalization of EMA, able to deal with masked implementation is introduced.
{"title":"The EVIL Machine: Encode, Visualize and Interpret the Leakage","authors":"Valence Cristiani, Maxime Lecomte, P. Maurine","doi":"10.1145/3555776.3577688","DOIUrl":"https://doi.org/10.1145/3555776.3577688","url":null,"abstract":"Unsupervised side-channel attacks allow extracting secret keys manipulated by cryptographic primitives through leakages of their physical implementations. As opposed to supervised attacks, they do not require a preliminary profiling of the target, constituting a broader threat since they imply weaker assumptions on the adversary model. Their downside is their requirement for some a priori knowledge on the leakage model of the device. On one hand, stochastic attacks such as the Linear Regression Analysis (LRA) allow for a flexible a priori, but are mostly limited to a univariate treatment of the traces. On the other hand, model-based attacks require an explicit formulation of the leakage model but have recently been extended to multidimensional versions allowing to benefit from the potential of Deep Learning (DL) techniques. The EVIL Machine Attack (EMA), introduced in this paper, aims at taking the best of both worlds. Inspired by generative adversarial networks, its architecture is able to recover a representation of the leakage model, which is then turned into a key distinguisher allowing flexible a priori. In addition, state-of-the-art DL techniques require 256 network trainings to conduct the attack. EMA requires only one, scaling down the time complexity of such attacks by a considerable factor. Simulations and real experiments show that EMA is applicable in cases where the adversary has very low knowledge on the leakage model, while significantly reducing the required number of traces compared to a classical LRA. Eventually, a generalization of EMA, able to deal with masked implementation is introduced.","PeriodicalId":42971,"journal":{"name":"Applied Computing Review","volume":"144 1","pages":""},"PeriodicalIF":1.0,"publicationDate":"2023-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77580079","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：