This AI Newsletter is all you need #16

What happened this week in AI by Louis

Diffusions are being used for everything: images, videos, 3D models and soon even audio thanks to StabilityAI, the company behind Stable Diffusion. StabilityAI announced an “AI Music Generator'' called Harmonai based on the “Dance Diffusion” Model, which is basically a stable diffusion model for generating sound instead of images. Diffusion-based models are definitely replacing convolutions and vision transformers even though they are using some of the same principles. We believe they will be here for a bit and are definitely worth learning more about, as they are just getting started and the field is still in the first experimentation phase with this architecture. Our friends at Delta Academy with whom we are partnering, might be building a complete and really innovative course on diffusion models. It would allow you to learn about the models and implement them by competing in game competitions with your peers. If that sounds interesting, stay tuned because we will make the announcements on the Discord channel! If you’d like to learn more about diffusion models, have a look at this article on Towards AI.

Hottest News

1. Deepmind tackles algorithm discovery with AlphaTensor. As I mention in the second paper below, Deepmind published “Discovering faster matrix multiplication algorithms with reinforcement learning” in which they tackle the task of discovering new algorithms using deep reinforcement learning to allow for faster and unique discoveries of new algorithms. In their blog post, they mention that it is the first extension of AlphaZero applied to mathematics already unlocking new possibilities for research! Read more in their blog post.
2. “Blueprint for an AI Bill of Rights.” The White House Office of Science & Technology Policy released a list of suggestions that developers, businesses, users, and lawmakers can follow to reduce AI’s potential harms! It has a lot of details covering all the what, why, and how behind the suggestions. Have a look!
3. Activeloop releases Deep Lake, the data lake for deep learning. Deep Lake stores complex data, such as images, videos, annotations, embeddings, and tabular data in the form of tensors and rapidly streams the data over the network to Tensor Query Language, in-browser visualization engine, and deep learning frameworks without sacrificing GPU utilization. Deep Lake outperforms all open-source data loaders (Webdataset, PyTorch, Torchdata, & others) on the landscape in a variety of scenarios. Read the announcement post, download the whitepaper, or watch the demo.

Most interesting papers of the week

1. Protein structure generation via folding diffusion. A new diffusion-based generative model that designs protein backbone structures via a procedure that mirrors the native folding process (“We describe protein backbone structure as a series of consecutive angles capturing the relative orientation of the constituent amino acid residues, and generate new structures by denoising from a random, unfolded state towards a stable folded structure.”).
2. Discovering faster matrix multiplication algorithms with reinforcement learning. “We report a deep reinforcement learning approach based on AlphaZero for discovering efficient and provably correct algorithms for the multiplication of arbitrary matrices.”
3. NOVEL VIEW SYNTHESIS WITH DIFFUSION MODELS 3DiM: a diffusion model for 3D novel view synthesis from as few as a single image.

Enjoy these papers and news summaries? Get a daily recap in your inbox!

The Learn AI Together Community section!

Meme of the week!

Meme once again shared by friedliver#0614. Share your meme with us on the meme channel!

Featured Community post from the Discord

An E-commerce text classification notebook!

This notebook by Sugata Ghosh, one of our members, goes over e-commerce text classification using TF-IDF and Word2Vec. It is thorough, well explained and very professional. A great resource for product classification and playing with text data. Check it out and let Sugata know your thoughts and feedback!

AI poll of the week!

Join the discussion on Discord.

TAI Curated section

Article of the week

Building Feedforward Neural Networks from Scratch by Nicolò Tognoni

Deep learning techniques are based on neural networks, commonly referred to as Artificial Neural Networks (ANNs), which are a subset of machine learning. You will learn more about Feed Forward Neural Networks in this article, starting with the fundamentals such as what a perceptron is and working your way up to backpropagation. With the use of interactive pictures, the author skillfully addressed all the crucial points while keeping the article's length in mind.

Our must-read articles

Adaptive Learning for Time Series Forecasting by Reza Yazdanfar

Building an End-to-End Linked Data Engineering Project by Edoardo Bianchi

If you are interested in publishing with Towards AI, check our guidelines and sign up. We will publish your work to our network if it meets our editorial policies and standards.

Job offers

Research Scientist in ML for Climate Modeling @ The Allen Institute for AI (Hybrid Remote)

Senior AI Software Engineer @ Spot AI (Remote)

Senior Machine Learning Researcher for Copilot @ Github (Remote)

Senior Software Engineer @ Captur (Remote, +/- 2 hours UK time)

Machine Learning Apprentice @ HingeHealth (Remote)

Senior ML Ops Engineer @ BenchSi (Remote)

ML Research Intern @ Genesis Therapeutics (Burlingame, CA)

Interested in sharing a job opportunity here? Contact [email protected] or post the opportunity in our #hiring channel on discord!

If you are preparing your next machine learning interview, don’t hesitate to check out our leading interview preparation website, confetti!