Original article was published by Jesus Rodriguez on Artificial Intelligence on Medium
The Sequence Scope: A Heroku for Machine Learning
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: A Heroku for Machine Learning
Years ago, cloud startup Heroku was able to successfully challenge cloud providers such as Amazon and Microsoft by providing a super simple model for building cloud applications. For years, Heroku was able to remain competitive in the midst of AWS and Azure growth, until it was acquired by Salesforce for $212 million. These days, it feels like we need a Heroku for the machine learning (ML) space.
A few weeks ago, we wrote about how big tech incumbents such as AWS, Microsoft and Google are vastly dominating the market for machine learning infrastructure technology. The same argument might suggest that it’s going to be difficult for startups in the ML infrastructure space to gain any traction when competing with these incumbents. However, that argument is not entirely correct. Building end-to-end ML applications in platforms such as AWS SageMaker or Azure ML still requires a lot of moving parts and interaction with “heavy” infrastructure components such as containers, etc. The market has a clear opportunity for startups that provide a simple model for building ML applications without caring about the underlying infrastructure. Just this week, the creators of the famous PyTorch Lightning framework launched Grid.ai to pursue that vision. Just like Heroku had its moment in the cloud infrastructure space, platforms like Grid.ai can fill a very obvious gap in the ML infrastructure market.
What do you think, who can become the new Heroku for machine learning?
Leave a comment
🔺🔻TheSequence Scope — our Sunday edition with the industry’s development overview — is free. To receive high-quality educational content every Tuesday and Thursday, please subscribe to TheSequence Edge 🔺🔻
🗓 Next week in TheSequence Edge:
Edge#29:privacy-preserving machine learning; Google’s PATE method for scalable private machine learning; the PySyft open-source framework for private deep learning.
Edge#30: the concept of active learning; how Amazon Research uses active learning to teach Alexa new skills; modAl — a modular framework for active learning.
Now, let’s review the most important developments in the AI industry this week.
🔎 ML Research
Berkeley AI Research (BAIR) Lab published a paper proposing a self-supervised visual reinforcement learning method that learns by discovering novelties in an environment ->read more BAIR Lab blog
Inside LinkedIn’s Recommendation System
The LinkedIn engineering team published an insightful blog post about the machine learning techniques used in its recommendation systems ->read more on the LinkedIn engineering team blog
Amazon Research published a paper proposing an advanced method for product recommendations ->read more on Amazon Research blog
🤖 Cool AI Tech Releases
Uber announced the release of the new version of Ludwig, its no-code machine learning platform ->read more on the Uber engineering team blog
Facebook AI Research (FAIR) open-sourced GTN, a framework for automatic differentiation in graph data structures ->read more on FAIR blog
Nvidia released a universal PyTorch library with an optimized implementation of various GAN images and video synthesis. It covers three types of models and provides tutorials for them ->read more on Github
Another release by Nvidia. They announced a platform that allows engineers to collaborate virtually and in real-time photorealistic simulation ->read more on Nvidia website
💸 Money in AI
- Cloud communication platform MessageBird has raised $200 million in a Series C round. MessageBird’s contact center software (chatbots and FAQ bots) is powered by AI and automation to detect sentiment and intent, translate between languages, automate a portion of customer queries.
- Payment automation solutions provider Tipalti raised $150 million in a Series E round. It claims that every step of the payables process — including self-service supplier management, data entry, form validations, fraud and regulatory controls, invoice processing and approvals, global payments execution, and payment reconciliation — is backed with integrated AI&ML. The company’s evaluation is over $2 billion in valuation.
- Unified communications platform Dialpad has raised $100 million in a Series E funding round. It actively leverages AI solutions for different tasks, such as using NLP to flag all the actionable items from meetings as well as carrying out sentiment analysis from inbound calls. The company’s evaluation is $1.2 billion.
- Data lineage automation platform Manta closed a $13 million Series A1 round. It helps IT and business professionals track data journeys, visualize its lineage, get actionable intelligence, and reduce costs on data management.
- Grid AI, a startup that helps ML engineers work more efficiently, has raised an $18.6 million Series A funding round. Grid AI was founded by William Falcon, the inventor of the famous open-sourced PyTorch Lightning project. Now, he is turning PyTorch Lightning into the core of Grid AI’s service, decoupling data science from engineering.
- AI-powered videoconferencing startup Headroom is coming out of stealth with a $5 million raised in a seed round. The startup leverages such AI tools as computer vision, natural language processing, and others to improve connections, optimize video quality and provide clients with transcripts, summaries, and other conversation analytics.
- Dental robotics startup Neocis raised $72 million in the latest round of funding. Except for the robotics part, Neocis’s software also leverages AI to make surgeons work easier, by providing warnings about proximity to the nerve or sinus and displaying information about the bone density quality in the proposed implant location. Machine learning is also used to analyze submitted data and smartly plan further procedures.
- Fitness startup Exer Labs announced $2 million in funding. Like other blooming sports apps, Exer Studio uses AI and computer vision to analyze trainees’ movements, their form, creating real-time audio feedback, workout summary, and a special motivational leaderboard.