From a697ab576abe11661c138efdd8606f769f3dda40 Mon Sep 17 00:00:00 2001 From: Sefik Ilkin Serengil Date: Mon, 10 Mar 2025 15:07:56 +0000 Subject: [PATCH] Update README.md --- README.md | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-) diff --git a/README.md b/README.md index 09a3af7..6fdc14e 100644 --- a/README.md +++ b/README.md @@ -374,9 +374,9 @@ If your task requires facial recognition on large datasets, you should combine D Conversely, if your task involves facial recognition on small to moderate-sized databases, you can adopt use relational databases such as [Postgres](https://youtu.be/f41sLxn1c0k) or [SQLite](https://youtu.be/_1ShBeWToPg), or NoSQL databases like [Mongo](https://youtu.be/dmprgum9Xu8), [Redis](https://youtu.be/X7DSpUMVTsw) or [Cassandra](https://youtu.be/J_yXpc3Y8Ec) to perform exact nearest neighbor search. -**Encrypt Embeddings** - [`Demo with PHE`](https://youtu.be/8VCu39jFZ7k), [`Demo with FHE`](https://youtu.be/njjw0PEhH00), [`Tutorial for PHE`](https://sefiks.com/2025/03/04/vector-similarity-search-with-partially-homomorphic-encryption-in-python/), [`Tutorial for FHE`](https://sefiks.com/2021/12/01/homomorphic-facial-recognition-with-tenseal/) +**Encrypt Embeddings** - [`Demo with PHE`](https://youtu.be/8VCu39jFZ7k), [`Tutorial for PHE`](https://sefiks.com/2025/03/04/vector-similarity-search-with-partially-homomorphic-encryption-in-python/), [`Demo with FHE`](https://youtu.be/njjw0PEhH00), [`Tutorial for FHE`](https://sefiks.com/2021/12/01/homomorphic-facial-recognition-with-tenseal/) -Even though vector embeddings are not reversible to original images, they still contain sensitive information such as fingerprints, making their security critical. Encrypting embeddings is essential for higher security applications to prevent adversarial attacks that could manipulate or extract sensitive information. Traditional encryption methods like AES are very safe but limited in securely utilizing cloud computational power for distance calculations. Herein, [homomorphic encryption](https://youtu.be/3ejI0zNPMEQ), allowing calculations on encrypted data, offers a robust alternative. In summary, we are able to compute encrypted similarity between encrypted embeddings with homomorphic encryption. +Even though vector embeddings are not reversible to original images, they still contain sensitive information such as fingerprints, making their security critical. Encrypting embeddings is essential for higher security applications to prevent adversarial attacks that could manipulate or extract sensitive information. Traditional encryption methods like AES are very safe but limited in securely utilizing cloud computational power for distance calculations. Herein, [homomorphic encryption](https://youtu.be/3ejI0zNPMEQ), allowing calculations on encrypted data, offers a robust alternative. ```python from lightphe import LightPHE @@ -413,11 +413,7 @@ calculated_similarity = cs.decrypt(encrypted_cosine_similarity)[0] print("same person" if calculated_similarity >= 1 - threshold else "different persons") ``` -In this scheme, we leverage the computational power of the cloud to compute encrypted cosine similarity. However, the cloud has no knowledge of the actual calculations it performs. Only the secret key holder on the on-premises side can decrypt the encrypted cosine similarity and determine whether the pair represents the same person or different individuals. - -Check out [`LightPHE`](https://github.com/serengil/LightPHE) library to find out more about partially homomorphic encryption. - -Additionally, you can opt for fully homomorphic encryption (FHE) instead of partially homomorphic encryption (PHE). However, FHE has certain limitations, including larger ciphertexts and keys, higher computational demands, and unsuitability for memory-constrained environments. Nevertheless, if you are determined to use FHE over PHE, you may consider exploring the [`CipherFace`](https://github.com/serengil/cipherface) library. It integrates DeepFace and TenSEAL, offering a simple interface for encrypting vector embeddings using FHE. +In this scheme, we leverage the computational power of the cloud to compute encrypted cosine similarity. However, the cloud has no knowledge of the actual calculations it performs. That's the magic of homomorphic encryption! Only the secret key holder on the on-premises side can decrypt the encrypted cosine similarity and determine whether the pair represents the same person or different individuals. Check out [`LightPHE`](https://github.com/serengil/LightPHE) library to find out more about partially homomorphic encryption. ## Contribution