Image by Darkmoon_Art from Pixabay

Our vision at Mithril Security

These last years we have seen the enormous progress made by AI algorithms. Computer vision has been dominated these last years by deep learning, showing excellent performances on very diverse tasks ranging from radiography analysis to biometric identification. We have also witnessed the emergence of smart assistants using voice recognition, helping us in our everyday life and in our homes.

While these results are quite promising, one question remains : can we trust these AI models ? …


Image by Darkmoon_Art from Pixabay

Confidential computing explained:

Part 1 : introduction

Part 2 : attestation

I — Introduction

We saw in the previous article why we need techniques such as confidential computing to secure sensitive workloads, and how different it is from regular methods.


Source: Pixabay

EDIT: This post is now available on OpenMined’s blog. Due to issues to write code and maths, I have decided to release the rest of the series on OpenMined for a more comfortable reader experience. Links are available here:

CKKS Explained: Part 1, Vanilla Encoding and Decoding

CKKS Explained: Part 2, Full Encoding and Decoding

CKKS Explained: Part 3, Encryption and Decryption

Homomorphic encryption intro:

Part 1: Overview and use cases

Part 2: HE landscape and CKKS

Part 3: Encoding and decoding in CKKS

Introduction

In the previous article, we saw what is Homomorphic Encryption, how it works, and had…


Description of the Homomorphic Encryption landscape, what is HE and a first look at CKKS for an HE scheme on complex numbers.

Source: Wikimedia

Homomorphic encryption intro:

Part 1: Overview and use cases

Part 2: HE landscape and CKKS

Part 3: Encoding and decoding in CKKS

I. Introduction

In the previous article https://medium.com/@dhuynh95/homomorphic-encryption-intro-part-1-overview-and-use-cases-a601adcff06c, we saw why we needed privacy preserving Machine Learning, how Homomorphic Encryption could achieve it and what use cases it could address.

In this article we will cover the basics of Homomorphic Encryption, and have a first look at the mechanics of one HE scheme, CKKS from the paper “Homomorphic Encryption for Arithmetic of Approximate Numbers” , which allows approximate arithmetic on real numbers, compared to other schemes which only work…


Source: Broadcom

Homomorphic encryption intro:

Part 1: Overview and use cases

Part 2: HE landscape and CKKS

Part 3: Encoding and decoding in CKKS

Introduction

Advancements in machine learning algorithms have resulted in a widespread adoption across industries. Nonetheless, areas dealing with sensitive and private data, like healthcare or finance, have lagged behind due to regulatory constraints to protect users’ data.

With the emergence of Machine Learning as a Service, entities are providing model inference as a service. We can distinguish three parties in such scenarios, a model owner such as a hospital who has trained a model, a host such as…


The hidden truth behind ML

Introduction

This article is also available on my blog https://danywind.github.io/2020/01/28/fast-neptune.html

Reproducibility has been an isssue in Machine Learning for some time. Not only does it pose a problem for research, as results shown in papers are often hard to reproduce in reality, but also in practice, where one wants to have the most robust pipeline, and reduce as much as possible the randomness when deploying a model. If ever one needs to retrain the model, for instance because it became obsolete, or minor changes have been added, it is real gain to know how the former best model was obtained.


Bobby Axelrod speaks the words!

Today, Neural Networks have made the headlines in many fields, such as image classification of cancer tissues, text generation, or even credit scoring. Nevertheless, one big issue that is rarely tackled with these models, is the uncertainty of the prediction.

When we humans learn, one of our greatest strengths is knowing our weaknesses, and not acting when there is too much uncertainty. However, the same is not true for most machine learning models, where decisions are taken without taking into account the uncertainty.

For instance, if you train a classifier on cats and dogs, it will only be able to…


Once upon a Tweet, I came across this conversation from Jeremy Howard quoting Yann LeCun about batch sizes :

Batch size discussion on Twitter

As this subject was something that has always been in some part of my mind since I came across the very nice learning rate finder from Fastai, I always wondered if there could be a useful batch size finder, that people could use to quickly start training their model with a good batch size.

As a reminder, the learning rate finder used in Fastai helps to find the right learning rate by testing different learning rates to find which one gives…


Introduction

As we all know, Deep Learning has really soared these past years, and has become the face of “AI” today. But both from a theoritical and practical point of view, can we really call “intelligent” an algorithm which not only needs millions of images, but also to label images, to understand what a cat is ?

We know that what makes humans so efficient at learning is to be able to learn from little data, create abstractions and act on it on several tasks. …

Daniel Huynh

CEO at Mithril Security

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store