In this paper, a group signature scheme based on quantum walk for quantum messages is proposed. Our scheme uses long step quantum walk-based teleportation and modified quantum one-time pad to authenticate the quantum messages respectively. In our scheme, the signer in a group signs the quantum messages by quantum walk-based teleportation and modified quantum one-time pad. The verifier can verify the signature via quantum walk-based teleportation while the group manager verifies the signature and identifies the signer via modified quantum one-time pad. The security analysis shows that the scheme can reach the properties of group signature. Compared to the teleportation via EPR pairs or Bell-like states, quantum walks are more flexible and use less measurement resources. Quantum teleportation via long-step quantum walk is more secure than that via one-step quantum walk in previous related works. The scheme can apply to arbitrary finite dimensional quantum systems and can also be possible to realize in practice.
{"title":"A Quantum Group Signature Based on Quantum Walk in d Dimensions","authors":"Yunxiao Qian, Haoyang Yu","doi":"10.1145/3546000.3546012","DOIUrl":"https://doi.org/10.1145/3546000.3546012","url":null,"abstract":"In this paper, a group signature scheme based on quantum walk for quantum messages is proposed. Our scheme uses long step quantum walk-based teleportation and modified quantum one-time pad to authenticate the quantum messages respectively. In our scheme, the signer in a group signs the quantum messages by quantum walk-based teleportation and modified quantum one-time pad. The verifier can verify the signature via quantum walk-based teleportation while the group manager verifies the signature and identifies the signer via modified quantum one-time pad. The security analysis shows that the scheme can reach the properties of group signature. Compared to the teleportation via EPR pairs or Bell-like states, quantum walks are more flexible and use less measurement resources. Quantum teleportation via long-step quantum walk is more secure than that via one-step quantum walk in previous related works. The scheme can apply to arbitrary finite dimensional quantum systems and can also be possible to realize in practice.","PeriodicalId":196955,"journal":{"name":"Proceedings of the 6th International Conference on High Performance Compilation, Computing and Communications","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133853688","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 ability to integrate information reaching us is a fundamental requirement for forming a coherent mental representation of our environment. One mechanism that has been proposed to underlie multisensory information across distributed cortical networks is transient synchronization of neural oscillations. Multisensory integration is a complex information processing, which is modulated by attention. In this study, we intended to explore the modulation of attention on neural oscillation of theta frequency band in audiovisual integration, by manipulating active attention to both visual and auditory stimuli or not attended at all. We analyzed the power of theta band, degree and long-range connectivity strength of functional brain networks in theta band. Our results showed that there was a significant difference in the power of theta frequency band between attended and unattended audiovisual integration, and the output degree of prefrontal area in attended theta network is significant higher than that in unattended network. Moreover, the strength of long-range connectivity from frontal area to parieto-occipital area is also significant higher in attended theta network of audiovisual integration, comparing to that in unattended theta network. We speculated that the top-down attention modulates the audiovisual integration, by increasing the neural oscillation of theta band, and that the top-down attention transmits theta signals to other regions through the frontal region, guiding other regions to integrate the visual and auditory inputs consciously.
{"title":"Attention Modulates the Neural Oscillation of Theta Frequency in Audiovisual Integration","authors":"Wenjing Wang, Guoao Liu, Yang Xi","doi":"10.1145/3546000.3546025","DOIUrl":"https://doi.org/10.1145/3546000.3546025","url":null,"abstract":"The ability to integrate information reaching us is a fundamental requirement for forming a coherent mental representation of our environment. One mechanism that has been proposed to underlie multisensory information across distributed cortical networks is transient synchronization of neural oscillations. Multisensory integration is a complex information processing, which is modulated by attention. In this study, we intended to explore the modulation of attention on neural oscillation of theta frequency band in audiovisual integration, by manipulating active attention to both visual and auditory stimuli or not attended at all. We analyzed the power of theta band, degree and long-range connectivity strength of functional brain networks in theta band. Our results showed that there was a significant difference in the power of theta frequency band between attended and unattended audiovisual integration, and the output degree of prefrontal area in attended theta network is significant higher than that in unattended network. Moreover, the strength of long-range connectivity from frontal area to parieto-occipital area is also significant higher in attended theta network of audiovisual integration, comparing to that in unattended theta network. We speculated that the top-down attention modulates the audiovisual integration, by increasing the neural oscillation of theta band, and that the top-down attention transmits theta signals to other regions through the frontal region, guiding other regions to integrate the visual and auditory inputs consciously.","PeriodicalId":196955,"journal":{"name":"Proceedings of the 6th International Conference on High Performance Compilation, Computing and Communications","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114746626","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}
Distributed deep learning is becoming increasingly popular due to the expanding demand for computing resources for deep learning models with a larger amount of parameters. Different from traditional training approaches, data-parallel training allows multiple compute nodes to train large deep learning models simultaneously in order to boost the training efficiency. In this paper, we present and compare six strategies for data-parallel training using PyTorch on the language model GPT-2 with 100M parameters using a qualitative approach. These strategies are Single GPU, Single Parameter Server, Distributed Parameter Server, Horovod, Distributed Parameter Server with Apex mixed-precision strategy, and Horovod with Apex mixed-precision strategy. We also analyze the quantitative experiment results from each strategy. In the end, we draw the conclusion that the Distributed Parameter Server with Apex mixed-precision strategy has the best performance on single node training, while Horovod with Apex is the most robust approach to use when we have single or multiple nodes.
{"title":"Modern Distributed Data-Parallel Large-Scale Pre-training Strategies For NLP models","authors":"Haoli Bai","doi":"10.1145/3546000.3546007","DOIUrl":"https://doi.org/10.1145/3546000.3546007","url":null,"abstract":"Distributed deep learning is becoming increasingly popular due to the expanding demand for computing resources for deep learning models with a larger amount of parameters. Different from traditional training approaches, data-parallel training allows multiple compute nodes to train large deep learning models simultaneously in order to boost the training efficiency. In this paper, we present and compare six strategies for data-parallel training using PyTorch on the language model GPT-2 with 100M parameters using a qualitative approach. These strategies are Single GPU, Single Parameter Server, Distributed Parameter Server, Horovod, Distributed Parameter Server with Apex mixed-precision strategy, and Horovod with Apex mixed-precision strategy. We also analyze the quantitative experiment results from each strategy. In the end, we draw the conclusion that the Distributed Parameter Server with Apex mixed-precision strategy has the best performance on single node training, while Horovod with Apex is the most robust approach to use when we have single or multiple nodes.","PeriodicalId":196955,"journal":{"name":"Proceedings of the 6th International Conference on High Performance Compilation, Computing and Communications","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2022-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129311371","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}
{"title":"Proceedings of the 6th International Conference on High Performance Compilation, Computing and Communications","authors":"","doi":"10.1145/3546000","DOIUrl":"https://doi.org/10.1145/3546000","url":null,"abstract":"","PeriodicalId":196955,"journal":{"name":"Proceedings of the 6th International Conference on High Performance Compilation, Computing and Communications","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122714500","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：