Proposal of security preserving machine learning of IoT
Bibliographic record
Abstract
The use of cloud computing system, which is the basic technology supporting ICT, is expanding. However, as the number of terminals connected to it increases, the limit of the capability is also becoming apparent. The limit of its capacity leads to the delay of significant processing time. As an architecture to improve this, the edge computing system has been proposed. This is known as a new paradigm corresponding the conventional cloud system. In the conventional cloud system, a terminal sends all data to the cloud and the cloud returns the result to the terminal or a thing directly connected to it. On the other hand, in the edge system, a plural of servers called edges are connected directly or to close distance between the cloud and the terminal (or thing). Then, let us consider the case of machine learning that requires big data. The purpose of learning is to find out the relationship (information) lurking in from the collected data. In order to realize this, a system with several parameters is assumed and estimated by repeatedly updating the parameters with learning data. Further, there is the problem of the security for learning data. In other words, users of cloud computing cannot escape the concern about the risk of information leakage. How can we build a cloud computing system to avoid such risks? Secure multiparty computation is known as one method of realizing safe computation. It is called SMC (Secure Multiparty Computation). Many studies on learning methods considering on SMC have also been proposed. Then, what kind of learning method is suitable for edge computing considering on SMC? In this paper, learning method suitable for edge computing considering on SMC is proposed. It is shown using an edge system composed of a client and m servers. Learning data are shared m pieces of subsets for m servers, learning is performed simultaneously in each server and system parameters are updated in the client using their results. The idea of learning method is shown using BP algorithm for neural network. The effectiveness is shown by numerical simulations.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| 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".