Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00013
Ken Asano, M. Natsui, T. Hanyu
This paper evaluates WKH bit-error tolerance of quantized neural networks (QNNs) for energy-efficient artificial intelligence (AI) applications utilizing stochastic properties of magnetic tunnel junction (MTJ) devices. Since QNNs have potentially high bit-error tolerance, they do not require large write currents to guarantee the certainty of the information held in the MTJ devices. By artificially adding bit errors to their weights, it is demonstrated that QNNs with binarized data representation achieve better error tolerance than any other ones in terms of the degradation rate of the recognition accuracy. In addition, based on the evaluation results, we show the possibility of reducing the write energy of MTJ devices up to 42% by exploiting high bit-error tolerance of the binarized QNN.
{"title":"Write-Energy Relaxation of MTJ-Based Quantized Neural-Network Hardware","authors":"Ken Asano, M. Natsui, T. Hanyu","doi":"10.1109/ISMVL57333.2023.00013","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00013","url":null,"abstract":"This paper evaluates WKH bit-error tolerance of quantized neural networks (QNNs) for energy-efficient artificial intelligence (AI) applications utilizing stochastic properties of magnetic tunnel junction (MTJ) devices. Since QNNs have potentially high bit-error tolerance, they do not require large write currents to guarantee the certainty of the information held in the MTJ devices. By artificially adding bit errors to their weights, it is demonstrated that QNNs with binarized data representation achieve better error tolerance than any other ones in terms of the degradation rate of the recognition accuracy. In addition, based on the evaluation results, we show the possibility of reducing the write energy of MTJ devices up to 42% by exploiting high bit-error tolerance of the binarized QNN.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114499389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00033
Hajime Machida
A centralizing monoid M is a set of unary operations which commute with some set F of operations. Here F is called a witness of M.On a 3-element set, majority minimal operations serve as witnesses of maximal centralizing monoids. In this paper, we start with one such majority operation and obtain its generalization, called mb, on a k-element set for any k ≥ 3. We explicitly describe the centralizing monoid M(mb) with mb as its witness and then prove it is not maximal if k > 3, contrary to the fact for k = 3.
{"title":"Search for Some Majority Operation and Studies of its Centralizing Monoid","authors":"Hajime Machida","doi":"10.1109/ISMVL57333.2023.00033","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00033","url":null,"abstract":"A centralizing monoid M is a set of unary operations which commute with some set F of operations. Here F is called a witness of M.On a 3-element set, majority minimal operations serve as witnesses of maximal centralizing monoids. In this paper, we start with one such majority operation and obtain its generalization, called mb, on a k-element set for any k ≥ 3. We explicitly describe the centralizing monoid M(mb) with mb as its witness and then prove it is not maximal if k > 3, contrary to the fact for k = 3.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121085471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00026
Bhavani Sampathkumar, Bailey Martin, Ritaja Das, P. Kalla, Florian Enescu
This paper introduces a novel concept of performing logic synthesis from multivariate polynomials with coefficients in the field of rationals $(mathbb{Q})$, where the variables take only Boolean values. Such polynomials are encountered during synthesis and verification of arithmetic circuits using computer algebra and algebraic geometry based techniques. The approach takes as input a polynomial f over $mathbb{Q}$ with binary variables, and derives a corresponding polynomial $tilde f$ over the finite field $left( {{mathbb{F}_2}} right)$ of two elements, such that f has the same variety (zero-set) as f. As ${mathbb{F}_2}$ is isomorphic to Boolean algebra, $tilde f$ can be translated to a Boolean network by mapping the products and sums as AND and XOR operators, respectively. We prove the correctness of our algebraic transformation, and present a recursive algorithm for the same. The translated $tilde f in {mathbb{F}_2}$ resultingly corresponds to a positive Davio decomposition, and is computed using both explicit and implicit representations. The approach is used to synthesize subfunctions of arithmetic circuits, under the partial synthesis framework. The efficacy of our approach is demonstrated over various integer multiplier architectures, where other contemporary approaches are infeasible.
本文介绍了在有理数$(mathbb{Q})$领域中，由变量仅取布尔值的多变量多项式进行逻辑综合的新概念。在使用计算机代数和基于代数几何的技术进行算术电路的综合和验证时，会遇到这样的多项式。该方法以具有二进制变量的多项式f / $mathbb{Q}$作为输入，并在两个元素的有限域$left( {{mathbb{F}_2}} right)$上导出相应的多项式$tilde f$，使得f与f具有相同的变量(零集)。由于${mathbb{F}_2}$与布尔代数同构，$tilde f$可以通过将乘积和和分别映射为and和异或运算符来转换为布尔网络。我们证明了代数变换的正确性，并给出了代数变换的递归算法。翻译后的$tilde f in {mathbb{F}_2}$结果对应于一个正的Davio分解，并使用显式和隐式表示进行计算。该方法在部分合成框架下用于合成算术电路的子函数。我们的方法的有效性在各种整数乘法器架构中得到了证明，而其他当代方法是不可行的。
{"title":"Logic Synthesis from Polynomials with Coefficients in the Field of Rationals","authors":"Bhavani Sampathkumar, Bailey Martin, Ritaja Das, P. Kalla, Florian Enescu","doi":"10.1109/ISMVL57333.2023.00026","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00026","url":null,"abstract":"This paper introduces a novel concept of performing logic synthesis from multivariate polynomials with coefficients in the field of rationals $(mathbb{Q})$, where the variables take only Boolean values. Such polynomials are encountered during synthesis and verification of arithmetic circuits using computer algebra and algebraic geometry based techniques. The approach takes as input a polynomial f over $mathbb{Q}$ with binary variables, and derives a corresponding polynomial $tilde f$ over the finite field $left( {{mathbb{F}_2}} right)$ of two elements, such that f has the same variety (zero-set) as f. As ${mathbb{F}_2}$ is isomorphic to Boolean algebra, $tilde f$ can be translated to a Boolean network by mapping the products and sums as AND and XOR operators, respectively. We prove the correctness of our algebraic transformation, and present a recursive algorithm for the same. The translated $tilde f in {mathbb{F}_2}$ resultingly corresponds to a positive Davio decomposition, and is computed using both explicit and implicit representations. The approach is used to synthesize subfunctions of arithmetic circuits, under the partial synthesis framework. The efficacy of our approach is demonstrated over various integer multiplier architectures, where other contemporary approaches are infeasible.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"234 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115502910","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00046
Tsutomu Sasao, A. Holmgren, P. Eklund
This paper analyzes data from coronary artery bypass grafting (CABG) using decision functions to represent rules. The data was collected at the University Hospital in Umeå, Sweden. The data contains pre-, intra-, and postoperative detail from 2975 heart operations during 1993-96. Each instance is represented by 14 preoperative variables, 4 intraoperative variables, and 9 postoperative variables. A logical method is used to predict the postoperative variables using preoperative variables. First, each postoperative variable is represented as a decision functions of preoperative variables. Then, for each postoperative variable, a minimal set of preoperative variables is derived. And finally, each postoperative variable is represented by a minimum set of rules using preoperative variables. With this method we can predict postoperative outcome, where prediction using preoperative data only is of particular interest e.g. for surgery scheduling.
{"title":"A Logical Method to Predict Outcomes After Coronary Artery Bypass Grafting","authors":"Tsutomu Sasao, A. Holmgren, P. Eklund","doi":"10.1109/ISMVL57333.2023.00046","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00046","url":null,"abstract":"This paper analyzes data from coronary artery bypass grafting (CABG) using decision functions to represent rules. The data was collected at the University Hospital in Umeå, Sweden. The data contains pre-, intra-, and postoperative detail from 2975 heart operations during 1993-96. Each instance is represented by 14 preoperative variables, 4 intraoperative variables, and 9 postoperative variables. A logical method is used to predict the postoperative variables using preoperative variables. First, each postoperative variable is represented as a decision functions of preoperative variables. Then, for each postoperative variable, a minimal set of preoperative variables is derived. And finally, each postoperative variable is represented by a minimum set of rules using preoperative variables. With this method we can predict postoperative outcome, where prediction using preoperative data only is of particular interest e.g. for surgery scheduling.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114072876","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ismvl57333.2023.00009
{"title":"Keynote I: Tackling the Explosions of Data and Solutions with Low-Bitwidth Computing Architectures","authors":"","doi":"10.1109/ismvl57333.2023.00009","DOIUrl":"https://doi.org/10.1109/ismvl57333.2023.00009","url":null,"abstract":"","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"171 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121372353","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00045
Rei Ueno, Yusuke Yagyu, N. Homma
This paper presents an efficient AES encryption/decryption hardware architecture with a fault detection scheme. The proposed hardware detects faults during encryption and decryption immediately to counter fault injection attacks such as differential fault analysis (DFA). The proposed hardware employs a fault detection scheme to verify the computation correctness and detect faults, where the datapath is divided into two sub-blocks of linear and nonlinear parts. An inverse function is executed one clock cycle after a half-of-round function. The hardware combines the proposed fault detection scheme with state-of-the-art datapath optimization techniques to achieve both efficient AES encryption/decryption and resistance to DFA. We showed through logic synthesis and simulation that the proposed AES hardware achieves 92% higher area-throughput efficiency and 47% lower power consumption than conventional hardware.
{"title":"Efficient DFA-Resistant AES Hardware Based on Concurrent Fault Detection Scheme","authors":"Rei Ueno, Yusuke Yagyu, N. Homma","doi":"10.1109/ISMVL57333.2023.00045","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00045","url":null,"abstract":"This paper presents an efficient AES encryption/decryption hardware architecture with a fault detection scheme. The proposed hardware detects faults during encryption and decryption immediately to counter fault injection attacks such as differential fault analysis (DFA). The proposed hardware employs a fault detection scheme to verify the computation correctness and detect faults, where the datapath is divided into two sub-blocks of linear and nonlinear parts. An inverse function is executed one clock cycle after a half-of-round function. The hardware combines the proposed fault detection scheme with state-of-the-art datapath optimization techniques to achieve both efficient AES encryption/decryption and resistance to DFA. We showed through logic synthesis and simulation that the proposed AES hardware achieves 92% higher area-throughput efficiency and 47% lower power consumption than conventional hardware.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128641205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00019
Yuki Sonetsuji, T. Isokawa, N. Kamiura, Hitoshi Tabuchi
In this paper, a method of detecting hypertensive subjects is proposed, using retinal fundus photographs classified by convolutional neural networks (CNNs for short). The proposed method employs Inception-v3 model as a CNN. The data to be presented to the proposed model are prepared from retinal fundus photographs. The scheme of fine tuning is conducted to construct a discrimination model. The model specifies the photographs corresponding to the subjects being in hypertensive states. Experimental results establish that the proposed method can achieves favorable metric values. Especially, Accuracy value achieved by the proposed method is comparatively high for the small scale of dataset, compared with the previously proposed method.
{"title":"On Neural-Network-Based Detection for Hypertensive Subjects Using Classification of Retinal Fundus Photographs","authors":"Yuki Sonetsuji, T. Isokawa, N. Kamiura, Hitoshi Tabuchi","doi":"10.1109/ISMVL57333.2023.00019","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00019","url":null,"abstract":"In this paper, a method of detecting hypertensive subjects is proposed, using retinal fundus photographs classified by convolutional neural networks (CNNs for short). The proposed method employs Inception-v3 model as a CNN. The data to be presented to the proposed model are prepared from retinal fundus photographs. The scheme of fine tuning is conducted to construct a discrimination model. The model specifies the photographs corresponding to the subjects being in hypertensive states. Experimental results establish that the proposed method can achieves favorable metric values. Especially, Accuracy value achieved by the proposed method is comparatively high for the small scale of dataset, compared with the previously proposed method.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131311376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00015
Kadir Emir, Jan Paseka
We show that for every orthogonal lub-complete poset P = (P,≤,′, 0, 1), we can introduce multiple-valued implications sharing properties with quantum implications presented for orthomodular lattices by Kalmbach. We call them classical implication, Kalmbach implication, and non-tolens implication.If the classical implication satisfies the order property, then the corresponding orthologic becomes classical and vice versa. If the Kalmbach or non-tolens implication meets the order property, then the corresponding orthologic becomes quantum and vice versa. A related result for the modus ponens rule is obtained.
{"title":"Kalmbach implication in orthomodular posets","authors":"Kadir Emir, Jan Paseka","doi":"10.1109/ISMVL57333.2023.00015","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00015","url":null,"abstract":"We show that for every orthogonal lub-complete poset P = (P,≤,′, 0, 1), we can introduce multiple-valued implications sharing properties with quantum implications presented for orthomodular lattices by Kalmbach. We call them classical implication, Kalmbach implication, and non-tolens implication.If the classical implication satisfies the order property, then the corresponding orthologic becomes classical and vice versa. If the Kalmbach or non-tolens implication meets the order property, then the corresponding orthologic becomes quantum and vice versa. A related result for the modus ponens rule is obtained.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125465682","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ISMVL57333.2023.00036
Thomas Grurl, Jürgen Fuß, R. Wille
By exploiting quantum mechanical effects, quantum computers can tackle problems that are infeasible for classical computers. At the same time, these quantum mechanical properties make handling quantum states exponentially hard—imposing major challenges on design tools. In the past, methods such as tensor networks or decision diagrams have shown that they can often keep those resource requirements in check by exploiting redundancies within the description of quantum states. But developments thus far focused on pure quantum states which do not provide a physically complete picture and, e.g., ignore frequently occurring noise effects. Density matrix representations provide such a complete picture, but are substantially larger. At the same time, they come with characteristics that allow for a more compact representation. In this work, we unveil this untapped potential and use it to provide a decision diagram representation that is optimized for density matrix representations. By this, we are providing a basis for more efficient design tools such as quantum circuit simulation which explicitly takes noise/error effects into account.
{"title":"Optimized Density Matrix Representations : Improving the Basis for Noise-Aware Quantum Circuit Design Tools","authors":"Thomas Grurl, Jürgen Fuß, R. Wille","doi":"10.1109/ISMVL57333.2023.00036","DOIUrl":"https://doi.org/10.1109/ISMVL57333.2023.00036","url":null,"abstract":"By exploiting quantum mechanical effects, quantum computers can tackle problems that are infeasible for classical computers. At the same time, these quantum mechanical properties make handling quantum states exponentially hard—imposing major challenges on design tools. In the past, methods such as tensor networks or decision diagrams have shown that they can often keep those resource requirements in check by exploiting redundancies within the description of quantum states. But developments thus far focused on pure quantum states which do not provide a physically complete picture and, e.g., ignore frequently occurring noise effects. Density matrix representations provide such a complete picture, but are substantially larger. At the same time, they come with characteristics that allow for a more compact representation. In this work, we unveil this untapped potential and use it to provide a decision diagram representation that is optimized for density matrix representations. By this, we are providing a basis for more efficient design tools such as quantum circuit simulation which explicitly takes noise/error effects into account.","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124912769","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01DOI: 10.1109/ismvl57333.2023.00010
{"title":"Keynote II: A Challenge of Scalable Quantum Computing Control Systems","authors":"","doi":"10.1109/ismvl57333.2023.00010","DOIUrl":"https://doi.org/10.1109/ismvl57333.2023.00010","url":null,"abstract":"","PeriodicalId":419220,"journal":{"name":"2023 IEEE 53rd International Symposium on Multiple-Valued Logic (ISMVL)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130214232","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：