CryptoKANs: Enhancing Privacy-Preserving Machine Learning in IoT Environments using Kolmogorov-Arnold Networks over Encrypted Data
Bibliographic record
Abstract
The proliferation of the Internet of Things (IoT) and the rapid advancement of Neural Networks (NNs) jointly facilitate the Internet of Artificially Intelligent Things. However, the shift towards cloud-based solutions raises significant privacy concerns, as sensitive data may be misused during NN inference tasks (predictions). To address these concerns, Homomorphic Encryption (HE) was introduced, allowing computations to be performed directly on encrypted data. Integrating NNs with HE presents challenges when introducing non-linearity into the model due to the constraints of current HE schemes that support only linear/polynomial functions. This often necessitates finding approximations for traditional activation functions (AFs) in NNs, leading to several issues: (i) Fixing an AF from the outset limits the model’s flexibility and may not yield the optimal fit for the data; (ii) Approximating a fixed AF requires significant effort to explore various approximation techniques and often leads to degradation in classification performance in exchange for reduced computational complexity; (iii) Avoiding certain approximations, if possible, to maintain accuracy increases the computational burden on the client side in the form of extra operations/communications. To overcome these challenges, we introduce CryptoKANs—Kolmogorov-Arnold Networks over Encrypted Data—a novel approach that enables NNs to operate over encrypted data by leveraging learnable AFs and symbolization to enhance privacy-preserving machine learning (PPML) for inference tasks. CryptoKANs allow the model to learn HE-suitable AFs as part of the training process. Experimental results demonstrate that CryptoKANs outperform traditional PPML models, achieving superior accuracy and, in some cases, even surpassing the performance of original models that operate on plaintext data. These findings underscore the potential of CryptoKANs to provide efficient, interpretable, accurate, and scalable private inference, marking a significant advancement in the field of PPML.
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 | 0.001 | 0.005 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".