Every week, my team at Invector Labs publishes a newsletter to track the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:
From the Editor: Data Privacy and AI
Data privacy is one of the biggest challenges of modern machine learning applications. In order to build machine learning models, researchers need to have complete access to datasets that often contain sensitive data. The idea of models that work effectively with encrypted datasets has long been an elusive goal of the machine learning space. While research in areas such as homomorphic encryption or secure, multi-party computation has been rapidly advancing, its adoption in machine learning stacks remains limited at best.
Incorporating data privacy techniques in mainstream deep learning frameworks is essential to unlock many scenarios in regulated industries or even mobile applications. In the past, we have seen companies such as DeepMind pioneer some efforts to build data privacy frameworks for deep learning solutions. This week, Facebook took an important step open sourcing a new project that brings data privacy capabilities to PyTorch. Certainly, in the next few years data privacy frameworks should become a common component of deep learning stacks.
Now let’s take a look at the core developments in AI research and technology this week:
Massively Multi-Lingual Machine Translation
Google Research published a paper proposing a new neural machine translation method that can scale massively across hundreds of languages.
An AI that can Answer Questions About Images
Microsoft Research published a paper about a new technique that can answer textual questions about large scenic images.
Inside LinkedIn Caption Generation Techniques
LinkedIn published a blog post detailing how they automatically generate captions for uploaded images.
Cool AI Tech Releases
Facebook open sourced the new version of PyTorch.
Data Privacy for Deep Learning
As part of the PyTorch release Facebook also open sourced CrypTen, a framework for native data privacy in deep learning solutions.
An Interpretability Framework for PyTorch
This week Facebook also open sourced Captum, an interpretability library for PyTorch.
AI in the Real World
Chinese AI Companies Blacklisted
The US-China trade war affected some to China’s top AI companies.
Fighting Deepfakes with Deepfakes
Tech giants are creating AIs that generate deepfakes in order to train detection models.
A National Facial Recognition System
India is creating a national facial recognition system to modernize the police force and criminal identification but there are some concerns about the increasing levels of surveillance.