The Sequence Scope: Quantum Machine Learning is Becoming Real
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. Give it a try by subscribing below:
(Core ML concepts + groundbreaking research papers and frameworks + AI news and trends) x 5 minutes, 3 times a week =…
📝 Editorial: Quantum Machine Learning is Becoming Real
The idea of combining quantum computing and machine learning has fascinated researchers for decades. The combination seems so powerful that there is an entire research body known as quantum machine learning dedicated to it. However, the ideas of quantum machine learning always felt purely theoretical and disconnected from real-world applications. That seems to be rapidly changing as quantum computing is becoming more available to mainstream researchers. Just this week, AWS unveiled Braket, a quantum computing infrastructure available as a native AWS service. The release follows a similar offer from Microsoft in the form of the Azure Quantum service.
As a discipline, quantum machine learning explores the idea of machine learning models that can run in quantum computing architectures. Fundamental mathematical principles of machine learning, such as linear algebra, take a different dimension in the quantum space so it’s only logical to think that machine learning models will need to be reimagined from the quantum world. Some deep learning frameworks have started to experiment with quantum machine learning. TensorFlow Quantum is one of the most interesting offerings in the space. Quantum machine learning is in a very nascent stage but we shouldn’t ignore the field. The important thing about quantum machine learning is that the first players to capture that market could have a massive differentiation over their competitors, as this type of model can perform at levels we haven’t seen before.
🗓 Next week in TheSequence Edge
Aug 18, Edge#13: the balance between model interpretability and accuracy; the famous “The Building Blocks of Interpretability” paper with a series of techniques to understand how neural networks make decisions; TensorBoard model visualization and interpretability toolset.
Aug 20, Edge#14: the concept of semi-supervised learning; a Google paper that proposes a data augmentation method to advance semi-supervised learning; Labelbox, a fast-growing platform for data labeling.
To stay up to date and receive TheSequence Edge every Tuesday and Thursday, please consider subscribing. Today is THE LAST DAY you can subscribe with a permanent 20% discount.
TheSequence Scope — our Sunday edition — is free.
Now, let’s review the most important developments in the AI industry this week.
🔎 ML Research
GANs and Transfer Learning
Microsoft Research published a paper exploring how adversarial robustness can improve transfer learning in computer vision models ->read more on Microsoft Research blog
Better Model Pretraining with Language Representations
Google Researchers published a paper proposing REALM, a technique that can improve language pre-trained models like BERT with explicit knowledge ->read more on Google AI blog
Challenges of Real-World Reinforcement Learning
A paper by Google Research outlining nine challenges that can hinder the application of reinforcement learning algorithms to real-world applications ->read more on Google AI blog
🤖 Cool AI Tech Releases
The intersection of machine learning and quantum computing is a new and fascinating area of research that just got a boost with the general availability announcement of AWS Braket, a native quantum computing infrastructure service ->read more on AWS blog
The Language Interpretability Tool (LIT)
Google-affiliated researchers developed an open-source platform for interactive visualization and understanding of NLP models. It offers an extensible and framework agnostic interface ->read more in the original paper ->code and full documentation are available on GitHub
💬 Useful Tweet
💸 Money in AI
- Revenue intelligence startup Gong raised a nice $200 million in Series D. Gong tries to create “revenue intelligence,” meaning they capture both sides of customer-salesperson interactions and then use artificial intelligence to transcribe and analyze those interactions. Sounds simple. With the new funding, they aim to hire 100 new people by the end of the year.
- AI-driven drug discovery platform Atomwise has raised $123 million in Series B financing. It invented the first deep learning AI technology for structure-based small molecule drug discovery. Its AI platform AtomNet® contains more than 16 billion molecules for virtual screening.
- AI-powered intranet platform Simpplr raised $10 million. The platform is designed to streamline communications. AI technologies ensure the content is relevant, up-to-date, and tailored to users.
- Email security startup Ironscales raised $8 million in its series B. Using AI tools, its self-learning email security platform quickly detects nefarious emails that slip through traditional anti-phishing defenses, responding to them automatically.
- AI cybersecurity startup Elisity raised $7.5 million in seed funding. Elisity Cognitive Trust combines zero-trust network access with an AI-enabled software-defined perimeter. It helps enterprises protect their data and assets while ensuring access without compromise to any application, data, or device.
If you find our newsletter useful, please consider supporting our efforts. Subscribe or make it a gift for those who can benefit from it. It’s the last week when you can get it with a permanent 20% discount.