The Sequence Scope: The AI Consolidation Movement will Continue in 2021
Weekly newsletter that discusses impactful ML research papers, cool tech releases, the money in AI, and real-life implementations.
The Sequence Scope is a summary of the most important published research papers, released technology and startup news in the AI ecosystem in the last week. This compendium is part of TheSequence newsletter. Data scientists, scholars, and developers from Microsoft Research, Intel Corporation, Linux Foundation AI, Google, Lockheed Martin, Cardiff University, Mellon College of Science, Warsaw University of Technology, Universitat Politècnica de València and other companies and universities are already subscribed to TheSequence.
📝 Editorial: The AI Consolidation Movement will Continue in 2021
Welcome to the last edition of The Sequence Scope in 2020. We started this educational newsletter a little over five months ago and today, we have more than 55,000 subscribers, including some of the top AI labs in the world. We really would like to thank you for your support and encouragement.
Well, this is the time of the year in which everyone starts making predictions about the next few months of a given technology trend. Don’t worry, we are not going to do that. Like the great New York Yankee’s catcher Yogi Berra once said, “It’s tough to make predictions, especially about the future”. However, there is one prediction that we feel pretty confident making and that is that the frenzy of AI acquisitions by the large technology firms will continue in 2021.
When we say frenzy, it’s not an exaggeration. The M&A engines in the AI space were incredibly active in 2020. Let’s do a quick recap of relevant M&A transactions in the AI space in 2020:
- Apple: Acquired companies like Xnor.ai, Voysis, Camerai and Scout FM.
- Microsoft: Acquired companies like ADRM Software and Orions Systems.
- Amazon: Acquired Zoox.
- Google: Acquired AppSheet.
- ServiceNow: Acquired ElementAI.
There are many more examples, but the ones illustrated above are certainly relevant. The rapid activity in AI M&A makes it a bit difficult for AI startups to achieve maturity, which represents a long-term challenge for the evolution of the market. However, it doesn’t seem this is going to change in 2021.
Now, let’s review the most important developments in the AI industry this week.
🔎 ML Research
Memorization and GPT-2
Researchers from the Berkeley AI Research Lab (BAIR) published an insightful paper about the security vulnerabilities created by the memorization of language pre-trained models ->read more on BAIR blog
Microsoft Research published an interesting paper about how to use “unadversarial examples” to increase the robustness of image classification models ->read more on Microsoft Research blog
More Optimal Transformers for Image Classification
Researchers from the Facebook AI Research (FAIR) and Sorbonne University published a paper proposing a method to optimize the training of image classification transformer models, which are notoriously data hungry ->read more in the original research paper
🤖 Cool AI Tech Releases
Inside Facebook’s Deep Learning Architecture
The Facebook AI Research (FAIR) team published an insightful blog post about the new architecture and frameworks they are using to improve the modularity and interoperability of their deep learning infrastructure ->read more on FAIR blog
💸 Money in AI
- Honeywell strengthens its position in AI and edge computing. It announced the acquisition of Sparta Systems for $1.3 billion to gain control of quality management software (QMS). Honeywell is going to augment it with machine learning algorithms and other AI capabilities for better decision making. At the same time, Honeywell Ventures is leading a $2 million round of seed funding for DarwinAI, a company that promises to make it simpler to build complex AI models using both machine and deep learning algorithms.
- Conversational AI intelligence startup DeepAffects was acquired by RingCentral under undisclosed financial terms. It’s part of RingCentral’s plan to create smarter video meetings. DeepAffects has a rich set of AI models and high accuracy for multi-speaker identification, emotion and speech recognition, and intent detection, which allows it to extract meaningful insights and build conversational metrics.
- Data analytics platform Second Measure was acquired by Bloomberg under undisclosed terms. The startup analyzes billions of anonymized purchases to help some of the world’s largest brands and investment firms answer real-time questions on company performance and consumer behavior. This should complement Bloomberg’s company fundamentals and market data, in the times of fast-growing demand for alternative data by investment firms that pursue more data-driven investment strategies.
- Fraud detection platform White Ops has been acquired by Goldman Sachs’ merchant banking division, ClearSky Security, and NightDragon under undisclosed terms. White Ops uses threat intelligence and machine learning to detect malicious impersonation of human behavior and combat automated bot-based attacks and frauds at scale.
- AI-powered office space analytics startup VergeSense raised $12 million in a Series B round of funding. VergeSense has developed proprietary deep-learning sensors and an analytics platform to give businesses insights into how physical spaces are being used, including counting people anonymously and detecting objects. Hiring.