检索结果-内蒙古大学图书馆

White Box Watermarking for Convolution Layers in Fine-Tuning Model Using the constant weight code

JOURNAL OF IMAGING 2023年第6期9卷 117-117页

作者： Kuribayashi, Minoru Yasui, Tatsuya Malik, Asad Okayama Univ Grad Sch Nat Sci & Technol Okayama 7008530 Japan Aligarh Muslim Univ Dept Comp Sci Aligarh 202001 India

Deep neural network (DNN) watermarking is a potential approach for protecting the intellectual property rights of DNN models. Similar to classical watermarking techniques for multimedia content, the requirements for DNN watermarking include capacity, robustness, transparency, and other factors. Studies have focused on robustness against retraining and fine-tuning. However, less important neurons in the DNN model may be pruned. Moreover, although the encoding approach renders DNN watermarking robust against pruning attacks, the watermark is assumed to be embedded only into the fully connected layer in the fine-tuning model. In this study, we extended the method such that the model can be applied to any convolution layer of the DNN model and designed a watermark detector based on a statistical analysis of the extracted weight parameters to evaluate whether the model is watermarked. Using a nonfungible token mitigates the overwriting of the watermark and enables checking when the DNN model with the watermark was created.

关键词： DNN watermark fine-tuning model constant weight code detection non-fungible token

来源：评论

学校读者我要写书评

暂无评论

coded DNN Watermark: Robustness against Pruning Models Using constant weight code

引用

JOURNAL OF IMAGING 2022年第6期8卷 152-152页

作者： Yasui, Tatsuya Tanaka, Takuro Malik, Asad Kuribayashi, Minoru Okayama Univ Grad Sch Nat Sci & Technol Okayama 7008530 Japan Aligarh Muslim Univ Dept Comp Sci Aligarh 202002 Uttar Pradesh India

Deep Neural Network (DNN) watermarking techniques are increasingly being used to protect the intellectual property of DNN models. Basically, DNN watermarking is a technique to insert side information into the DNN model without significantly degrading the performance of its original task. A pruning attack is a threat to DNN watermarking, wherein the less important neurons in the model are pruned to make it faster and more compact. As a result, removing the watermark from the DNN model is possible. This study investigates a channel coding approach to protect DNN watermarking against pruning attacks. The channel model differs completely from conventional models involving digital images. Determining the suitable encoding methods for DNN watermarking remains an open problem. Herein, we presented a novel encoding approach using constant weight codes to protect the DNN watermarking against pruning attacks. The experimental results confirmed that the robustness against pruning attacks could be controlled by carefully setting two thresholds for binary symbols in the codeword.

关键词： watermarking pruning attack DNN model constant weight code fine-tuning

来源：评论

学校读者我要写书评

暂无评论

Completely reducible super-simple (v, 4, 4)-BIBDs and related constant weight codes

引用

JOURNAL OF COMBINATORIAL DESIGNS 2025年第1期33卷 27-36页

作者： Chen, Jingyuan Yu, Huangsheng Abel, R. Julian R. Wu, Dianhua Guangxi Normal Univ Sch Math & Stat Guilin 541006 Peoples R China UNSW Sydney Sch Math & Stat Sydney NSW Australia

A design is said to be super-simple if the intersection of any two blocks has at most two elements. A design with index lambda is said to be completely reducible, if its blocks can be partitioned into nonempty collections B-i, 1 <= i <= lambda, such that each B-i together with the point set forms a design with index unity. In this paper, it is proved that there exists a completely reducible super-simple (CRSS) (v, 4, 4) balanced incomplete block design ((v, 4, 4)-BIBD for short) if and only if v >= 13 and v equivalent to 1 or 4 (mod 12). A q-ary constant weight code (CWC) of length v with weight w and distance d is denoted as a (v, d, w)(q) code. The maximum size of a (v, d, w)(q) code is denoted as A(q) (v, d, w)(q) and the (v, d, w)(q) codes achieving this size are called optimal. CRSS designs with index q - 1 are closely related to q-ary CWCs. By using the results of CRSS (v, 4, 4)-BIBDs, A(5)(v, 6, 4)s are determined for all v equivalent to 0, 1, 3, 4 (mod 12), v >= 12.

关键词： BIBD completely reducible constant weight code GDD super-simple

来源：评论

学校读者我要写书评

暂无评论

On optimal constant weight codes derived from ω-circulant balanced generalized weighing matrices

引用

DESIGNS codeS AND CRYPTOGRAPHY 2024年第10期92卷 2791-2799页

作者： Kharaghani, Hadi Pender, Thomas Tonchev, Vladimir Univ Lethbridge Dept Math & Comp Sci Lethbridge AB T1K 3M4 Canada Univ Simon Fraser Dept Math Burnaby BC V5A 1S6 Canada Michigan Technol Univ Math Sci Houghton MI USA

Balanced generalized weight matrices are used to construct optimal constant weight codes that are monomially inequivalent to codes derived from the classical simplex codes. What's more, these codes can be assumed to be generated entirely by omega-shifts of a single codeword where omega is a primitive element of a Galois field. Additional constant weight codes are derived by projecting onto subgroups of the alphabet sets. These too are shown to be optimal.

关键词： constant weight code Equidistant nonlinear code Inequivalent codes Optimal code omega-circulant balanced generalized weighing matrix

来源：评论

学校读者我要写书评

暂无评论

Quantum search algorithm for binary constant weight codes

引用

QUANTUM INFORMATION PROCESSING 2024年第10期23卷 1-27页

作者： Yukiyoshi, Kein Ishikawa, Naoki Yokohama Natl Univ Fac Engn Yokohama Kanagawa 2408501 Japan

A binary constant weight code is a type of error-correcting code with a wide range of applications. The problem of finding a binary constant weight code has long been studied as a combinatorial optimization problem in coding theory. In this paper, we propose a quantum search algorithm for binary constant weight codes. Specifically, the search problem is formulated as a polynomial binary optimization problem and Grover adaptive search is used for providing the quadratic speedup. Focusing on the inherent structure of the problem, we derive an upper bound on the minimum of the objective function value and a lower bound on the exact number of solutions. By exploiting these two bounds, we successfully reduced the constant overhead of the algorithm, although the overall query complexity remains exponential due to the NP-complete nature of the problem. In our algebraic analysis, it was found that this proposed algorithm is capable of reducing the number of required qubits, thus enhancing the feasibility. Additionally, our simulations demonstrated that it reduces the average number of classical iterations by 63% as well as the average number of total Grover rotations by 31%. The proposed approach may be useful for other quantum search algorithms and optimization problems.

关键词： Coding theory constant weight code Quantum computing Grover adaptive search

来源：评论

学校读者我要写书评

暂无评论

On The Maximum Cliques Of The Subgraphs Induced By Binary constant weight codes In Powers Of Hypercubes

引用

COMPUTER JOURNAL 2023年第10期66卷 2535-2541页

作者： Shi, Juanjuan Kou, Yongfang Hu, Yulan Yang, Weihua Taiyuan Univ Technol Dept Math Taiyuan 030024 Shanxi Peoples R China

The problem of finding the maximum independent sets (or maximum cliques) of a given graph is fundamental in graph theory and is also one of the most important in terms of the application of graph theory. Let A(n, d, w) be the size of the maximum independent set of Q(n)((d-1,w)), which is the induced subgraph of points of weight w of the d - 1(th)-power of n-dimensional hypercubes. In order to further understand and study the dependent set of Q(n)((d-1,w)), we explore its clique number and the structure of the maximum clique. This paper obtains the clique number and the structure of the maximum clique of Q(n)((d-1,w)) for 5 <= d <= 6. Moreover, the characterizations for A(n, d, w) = 2 and 3 are also given.

关键词： Hypercubes Maximum clique Maximum independent set constant weight code

来源：评论

学校读者我要写书评

暂无评论

On the Maximum Cliques in Powers of Hypercubes

引用

JOURNAL OF INTERCONNECTION NETWORKS 2025年第1期25卷

作者： Hu, Yanting Yang, Weihua Taiyuan Univ Technol Dept Math Taiyuan 030024 Shanxi Peoples R China

Finding the values of A(n,d) and A(n,d,w) is a fundamental problem in classical coding theory. The A(n,d,w) is the size of the maximum independent set of Q(n)((d-1,w)), which is the induced subgraph of vectors of weight w of the (d - 1)th-power of n-dimensional hypercubes. Obviously, A(n,d,w) = alpha(Q(n)((d-1,w))). In order to further understand and study the independent set of Q(n)((d-1,w)), we explore its clique number and the structure of the maximum clique. In this paper, we obtain the clique number and the structure of the maximum clique of Q(n)((d,w)) for 6 <= d <= 7. As an application, by alpha(G)w(G) <= |V (G)|, we obtain the upper bounds of A(n,d,w) for 7 <= d <= 8.

关键词： Maximum clique hypercubes constant weight code clique number

来源：评论

学校读者我要写书评

暂无评论

On a class of optimal constant weight ternary codes

引用

DESIGNS codeS AND CRYPTOGRAPHY 2023年第1期91卷 45-54页

作者： Kharaghani, Hadi Suda, Sho Zaitsev, Vlad Univ Lethbridge Dept Math & Comp Sci Lethbridge AB T1K 3M4 Canada Natl Def Acad Japan Dept Math Yokosuka Kanagawa 2398686 Japan

A weighing matrix W of order n = p(m+1)-1/p-1 and weight p(m) is constructed and shown that the rows of W and -W together form optimal constant weight ternary codes of length n, weight p(m) and minimum distance p(m-1) (p+3/2) for each odd prime power p and integer m >= 1 and thus A(3) (p(m+1)-1/p - 1, p(m-1)(p + 3/2), p(m)) = 2(p(m+1)-1/p - 1).

关键词： Weighing matrix Ternary code constant weight code Optimal code Johnson bound

来源：评论

学校读者我要写书评

暂无评论

Optimal binary constant weight codes and affine linear groups over finite fields

引用

DESIGNS codeS AND CRYPTOGRAPHY 2019年第8期87卷 1815-1838页

作者： Hou, Xiang-Dong Univ S Florida Dept Math & Stat Tampa FL 33620 USA

The affine linear group of degree one, AGL(1,Fq), over the finite field Fq, acts sharply two-transitively on Fq. Given S<AGL(1,Fq) and an integer k, 1kq, does there exist a k-element subset B of Fq whose set-wise s... 详细信息

The affine linear group of degree one, AGL(1,Fq), over the finite field Fq, acts sharply two-transitively on Fq. Given Sconstant weight codes that are constructed from the action of AGL(1,Fq) on Fq to meet the Johnson bound. Consequently, for many parameters, we are able to determine the values of the function A2(n,d,w), which is the maximum number of codewords in a binary constant weight code of length n, weight w and minimum distance d.

关键词： Affine linear group BIBD constant weight code Johnson bound

来源：评论

学校读者我要写书评

暂无评论

Dimming-Aware Deep Learning Approach for OOK-Based Visible Light Communication

引用

JOURNAL OF LIGHTWAVE TECHNOLOGY 2020年第20期38卷 5733-5742页

作者： Zou, Cong Yang, Fang Tsinghua Univ Dept Elect Engn Beijing Natl Res Ctr Informat Sci & Technol Beijing 100084 Peoples R China Tsinghua Univ Shenzhen Key Lab Digital TV Syst Guangdong Prov & Shenzhen Res Inst Shenzhen 518057 Peoples R China

Visible light communication (VLC) is a secure, low-cost, and high-rate communication method. On-off keying (OOK) is one of the modulation schemes of VLC, turning each light either on or off to generate binary signals. Recently, deep learning (DL) technologies have made a series of breakthroughs for dimming in VLC system. This task is actually quite challenging for DL, since the VLC system needs to be able to support various dimming targets on account of the different preferences from users in practical applications, resulting in an optimization problem with multiple constraints. This article presents a DL framework for the dimming-aware binary VLC system, which can meet arbitrary dimming requirements by a universal neural network, named universal auto-encoder (UAE). The proposed UAE creatively utilizes a multi-branch architecture with several carefully designed concatenated patches, and a novel multi-stage training strategy for the optimization problem with multiple dimming constraints. The experiments indicate that the proposed DL approach outperforms existing techniques in terms of the average bit error rate, the satisfaction of the dimming constraints, and the robustness for imperfect optical channels.

关键词： Encoding Visible light communication Training Light emitting diodes Optical pulses Maximum likelihood decoding Deep learning constant weight code constrained optimization deep learning on-off keying visible light communication

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：