Efficient and Secure Neural Network Inference with Homomorphic Encryption
Bibliographic record
Abstract
Running deep neural networks in the cloud raises privacy concerns. While privacy-preserving machine learning techniques offer promising solutions, they often suffer from long latency due to large polynomials' replication costs and computational complexity. Diagonal-order matrix multiplication presents a potential avenue for addressing this issue by accelerating ciphertext-to-plaintext operations by efficiently using rotations. However, further performance enhancements are required to make these methods feasible for various platforms. This paper introduces the sparsity tailored for diagonal-order ciphertext-toplaintext operations with different ratios. By removing several weights parallel to the main diagonal, computational operations, such as rotations, addition, and multiplication, are reduced. Our approach significantly speeds up computations and lowers encrypted inference latency while maintaining model accuracy. Experimental evaluations reveal that with a 1:4 diagonal sparsity ratio, the proposed approach achieves a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$1.86 \times$</tex> speedup compared to traditional fully connected networks. This method also reduces memory usage to less than 0.343 GByte for the entire homomorphic process on the MNIST dataset.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.003 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".