Privacy-Preserving Convolutional Neural Network Classification Scheme With Multiple Keys
Bibliographic record
Abstract
Convolutional Neural Networks (CNNs) possess extensive applicability across diverse domains, particularly in the realm of image recognition. In light of the advent of machine learning as a service, the utilization of a well-trained CNN model by servers to execute image classification based on user queries has become a significant service, catering to a wide array of applications. Nevertheless, this convenience is accompanied by the inherent risk of data privacy and model privacy disclosure, which can have severe ramifications, particularly in the context of specialized scenarios like medical images and location images. Hence, how to perform classification for CNN with privacy protection emerges as a crucial research concern. Furthermore, the nonlinearity of CNN's activation function renders it unsuitable for homomorphic cryptosystems. In order to address these challenges, we put forth a privacy-preserving CNN classification scheme employing a distributed two trapdoors public-key cryptosystem (DT-PKC). Initially, we introduce a security protocol toolkit encompassing protocols for secure multiplication, secure activation function computing, and average pooling. In addition, we propose a novel continuous and derivative Tanhplus function as an approximation of the Relu function, aiming to enhance the accuracy of classification results. The secure activation function computing protocol utilizes the aforementioned Tanhplus function in conjunction with the proposed homogenization algorithm to compute the activation function. This protocol guarantees more precise and accurate output in the activation function calculation of CNN when operating under ciphertext. Furthermore, the adoption of the DT-PKC cryptosystem not only ensures privacy protection for CNN classification but also provides support for lightweight users and multiple keys. Finally, security analysis and performance evaluations demonstrate that the proposed scheme is secure, practicable, and efficient with high accuracy.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.003 | 0.006 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".