Convolutional Neural Network Protection Method of Lenet-5-Like Structure

Sun, Lei; Wang, Yuehan; Dai, Leyu

doi:10.1145/3297156.3297224

Cited by 4 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Sun et al [48] showed-using the example of a LeNet-5 CNN-that adapting the activation function of the convolutional layers to be dependent on a random number only known to the legitimate user can provide effective protection.…”

Section: B Model Parameter Protectionmentioning

confidence: 99%

Identifying Appropriate Intellectual Property Protection Mechanisms for Machine Learning Models: A Systematization of Watermarking, Fingerprinting, Model Access, and Attacks

Lederer

Mayer

Rauber

2024

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

The commercial use of machine learning (ML) is spreading; at the same time, ML models are becoming more complex and more expensive to train, which makes intellectual property protection (IPP) of trained models a pressing issue. Unlike other domains that can build on a solid understanding of the threats, attacks, and defenses available to protect their IP, ML-related research in this regard is still very fragmented. This is also due to a missing unified view as well as a common taxonomy of these aspects. In this article, we systematize our findings on IPP in ML while focusing on threats and attacks identified and defenses proposed at the time of writing. We develop a comprehensive threat model for IP in ML, categorizing attacks and defenses within a unified and consolidated taxonomy, thus bridging research from both the ML and security communities.

show abstract

Section: B Model Parameter Protectionmentioning

confidence: 99%

Identifying Appropriate Intellectual Property Protection Mechanisms for Machine Learning Models: A Systematization of Watermarking, Fingerprinting, Model Access, and Attacks

Lederer

Mayer

Rauber

2024

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

show abstract

“…The reason for their popularity is they can be seen as multiple layers of approximation functions, i.e., f (X) = f l • ... • f 2 • f 1 (X), where f s are some set of "functions" that are comprised of different weighted multiplications and additions called a neuron. In particular, fully connected deep neural networks have become the most popular because of their success in areas such as image processing [36].…”

Section: Neural Accelerationmentioning

confidence: 99%

“…The second fact is that these are relatively simple models compared to other neural network models used in machine learning. For example, consider Lenet-5 [36] which is used for handwritten digit recognition (numbers 0-9) and is a very common framework to build other networks. If the same Lenet-5 framework was applied to this problem with sparse matrix input of size 128 (i.e., n=128 in place of current n=512), there would be approximately 6.8 million trainable perimeters.…”

Section: B Ordering Selectionmentioning

confidence: 99%

Neural Acceleration of Graph Based Utility Functions for Sparse Matrices

Booth

Bolet

2023

IEEE Access

View full text Add to dashboard Cite

Many graph-based algorithms in high performance computing (HPC) use approximate solutions due to having algorithms that are computationally expensive or serial in nature. Neural acceleration, i.e., the process of speeding up approximation computation elements via artificial neural networks, is relatively new and has not focused on HPC graph-based algorithms. In this paper, we propose a starting point for applying models for neural acceleration to graph-based HPC algorithms utilizing an understanding of the connectivity computational pattern, recursive neural networks, and graph neural networks. We demonstrate these techniques on the problem related to the utility functions of sparse matrix ordering and fill-in (i.e., zero elements becoming nonzero during factorization) calculations. The problem of sparse matrix ordering is commonly used for issues related to load balancing, improving memory reuse, or reducing computational and memory costs in direct sparse linear solver methods. These utility functions are ideal for demonstration as they comprise a number of different graph-based subproblems, and thus demonstrate the usefulness of our method over a wide range. We show that we can accurately approximate the best ordering and the nonzero count of the sparse factorization matrix while speeding up the calculation by as much as 30.3× over the traditional serial method.

show abstract