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
How to select the right machine learning model for a problem? That question hunts data scientists on every single machine learning scenario. Deciding which algorithms to use, how many layers or nodes in neural network is far from being an easy endeavor. Interestingly enough, this is the type of problem in which machine learning excels. Can we use machine learning to build better machine learning architectures?
Automated machine learning(AutoML) is a technique pioneered by Google to streamline the creation of machine learning models. Recently, researchers have been testing the ideas behind AutoML to help with model search and architecture. Today we have notable efforts from Microsoft, Waymo or Salesforce in this area. While data scientists rely on subjective knowledge to architect machine learning mdoels, AutoML techniques can empirically test thousands of machine learning variations for a given problem. While human expertise is certainly required, AutoML can reduce the efforts drastically.
Now let’s take a look at the core developments in AI research and technology this week:
Google reviews some of their major AI research efforts in 2018
Researchers from Stanford University proposed a new method for creating natural language agents that can train themselves real time.
The AI team at Alphabet’s subsidiary Waymo discusses how they use AutoML for creating better machine learning models.
Cool Tech Releases
LinkedIn discusses their business monitoring platform which heavily uses unsupervised learning techniques.
The Facebook AI Research(FAIR) team summarized some of their accomplishments last year.
Microsoft launched the Ability Initiative to improve image captioning for blind people
AI in the Real World
AI Researchers from Tel-Aviv University created a neural network capable of generating images from a food recipe without ever being trained on the final dish. Pretty amazing.
Amazon is launching a new conference dedicated to AI and Robotics.
Cyber security researchers are discovering new ways in which AI agents can be attacked.