Last Week in AI: Announcing The Sequence of AI Knowledge
A new AI education project I have been working on.
We are expanding our weekly newsletter to focus on AI education( research papers, technology frameworks, practical implementations). You can see the result at http://thesequence.ai. But please read below.
It’s been almost two years since we started This Week in AI at Invector Labs. During that time, our audience has grown to tens of thousands of loyal readers. The original goal of the newsletter was to try to keep our audience up to date with the hottest developments in AI research and technology of the previous week. We try to not focus on news but on actual research and technology releases.
With our work building AI solutions at Invector Labs and with this newsletter we have noticed that the challenges for AI education continue increasing. The volume of published AI research has more than quadrupled in recent years and high quality machine learning technologies pop up every week. More importantly, the gap between AI research and practical implementations keeps widening. Keeping up with that fast growing ecosystem is really hard and feels overwhelming to developers and technology professionals wanting to get into the space. You can take machine learning courses but those consume a lot of time and, frankly, don’t address the challenges of keeping up with research and technology developments.
To contribute in a small way to address these challenges, we are going to be expanding this newsletter with a new offering.
Welcome to The Sequence of AI Knowledge: http://thesequence.ai
We are expanding This Week in AI into two newsletters: The Sequence Edge and The Sequence Scope
The Sequence Edge, we will be delivered to your inbox twice a week. In 5 minutes, this new newsletter will cover three key topics:
- A machine learning concept you should learn( ex: Transformers, Auto-ML, meta-learning, etc.)
- A summary of a research paper that has had an impact in the machine learning space.
- A briefing on a new machine learning framework or platform.
The Sequence Edge will be a subscription offering of $5 per month. You can subscribe with a permanent 20% discount at https://thesequence.substack.com/ThisWeekInAI
The Sequence Scope will be the new brand for this newsletter and will continue bringing you a summary of the most important research, technology releases and AI news of the week. You will continue receiving The Sequence Scope at no cost.
Just to be clear, we are expanding (not replacing) This Week in AI. We would love for you to join us in this experiment to bring AI education to your inbox 5 minutes at a time by subscribing to The Sequence Edge.
But if you prefer not to, you will still receive The Sequence Scope every Sunday FOR FREE.
Thanks for your support and loyalty throughout these two years and I can’t wait to get your feedback about our next phase.
Now let’s take a look at the core developments in AI research and technology this week:
AI Research:
New Approach to Reinforcement Learning
Google Researh published a paper proposing a new reinforcement learning method that prevents agents to use rough approximations for their decisions.
>Read more in this blog post from Google Research
Language Queries Over Tabular Data
Facebook AI Research published a paper proposing TaBERT, a transformer model to query tabular data using natural language.
>Read more in this blog post from the Facebook Research team
AutoML-Zero
Google Research published a paper proposing AutoML-Zero, a method that can create entire machine learning models starting with high school math operations. We cover this paper at TheSequence.
>Read more in this blog post from the Google Research team
Cool AI Tech Releases:
Azure Cognitive Services
Microsoft added a new group of services and capabilities to the Azure Cognitive Services stack.
>Read more in this blog post from the Azure team
Google Cloud Data QnA
Google Cloud released a new service to query petabytes of data in BigQuery using natural language.
>Read more in this blog post from the BigQuery team
AI in the Real World:
No-Code Text Analytics Startups Raises $2.2M
MonkeyLearn, a startup that provides a no-code experience for training and creating language models, raised $2.2 seed round.
>Read more in this coverage from TechCrunch
Neuroscience Startup Raises $53M
Kernel, a startup building brain technology applicable to business scenarios, just announced a monster $53 million round.
>Read more in this coverage from VentureBeat
Amazon Spends $100M to Retain Zoox Talent
Amazon is acquiring self-driving startup Zoox for about $1.3 billion but, additionally, is planning to spend another $100 million in stock awards to retain the talent.
