Join thousands of AI enthusiasts and experts at the Learn AI Community.

Publication

Latest   Newsletter

This AI newsletter is all you need #23

Last Updated on November 29, 2022 by Editorial Team

Author(s): Towards AI Editorial Team

Originally published on Towards AI the World’s Leading AI and Technology News and Media Company. If you are building an AI-related product or service, we invite you to consider becoming an AI sponsor. At Towards AI, we help scale AI and technology startups. Let us help you unleash your technology to the masses.

What happened this week in AI by Louis

Stable Diffusion 2.0. Obviously!

Stable Diffusion V1 has been a major milestone for AI adoption in our view, both in terms of mainstream use and awareness, and in developer engagement (reaching 33k GitHub stars in less than 2 months). We hope this success encourages widespread adoption of this open source release model at AI companies and startups. We expect this model will also lead to faster iteration and improvement and Stable Diffusion is delivering on this with the rapid release of version 2.0.

This second version offers the same service with better results and more features! It uses new diffusion models in order to generate base images of 512×512 pixels and 768×768 pixels, which you can upscale using their super-resolution upscaler to 2048×2048 or even higher. They also updated their inpainting model and introduced a depth-to-image model that generates new images using both the text and depth information. It’s very exciting to see stable diffusion continuing to advance as quickly as it has, and is a sign that computer vision is thriving!

Hottest News

  1. CICERO: An AI agent that negotiates, persuades, and cooperates with people
    Meta AI just introduced CICERO — “the first AI to achieve human-level performance in Diplomacy, a strategy game which requires building trust, negotiating and cooperating with multiple players.” They also open-sourced the code for it!
  2. Google launches new search, shopping and maps features
    Google has announced a lot of cool new features for Google Search, Google Lens, Shopping, and Maps, pretty much all their apps. The most exciting ones are the modifications to Google Search for finding food and the improvement on the translation for Google Lens. It was honestly quite bad in my experience from Hebrew to English! You can even see shoes you want to buy with AR using Google Shopping. This is getting really cool. I’m excited to try those out! Read more about the changes here.
  3. ‘SuperGPS’ Technology Accurately Pinpoints Your Position Within Inches
    Researchers have come up with a new and improved technology called SuperGPS that is accurate within 10 centimeters (or 3.9 inches) and doesn’t rely on navigation satellite systems. “We realized that with a few cutting-edge innovations, the telecommunication network could be transformed into a very accurate alternative positioning system that is independent of GPS.”

Three 5-minute reads/videos to keep you learning

  1. Observing Information Propagation in Deep Neural Networks
    For this first week of the “5-minute reads”, I want to share a short article about some of the work I did during my Master’s degree at École de Technologie Supérieure. We share an analogy with light and colors to better explain deep neural networks 🙂
    More to come in a conference near you soon! Let me know what you think!
  2. Neural Networks Demystified by Welch Labs
    This is a great YouTube series of short videos explaining neural networks, forward propagation gradient descent, and more. It is an amazing introduction video or refresher about neural networks! Amazing animations, super clear, code, math, everything you need to be entertained and learn in less than 5 minutes per video!
  3. What is Machine Learning? by Roberto Iriondo
    “Learn what machine learning (ML) is, how it works, and its importance in five minutes.” How convenient for our newsletter! As Roberto, one of my friends and co-founder at Towards AI mentions, the article is for anyone curious who wants a straightforward and accurate overview of what machine learning is, how it works, and its importance.

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

The Learn AI Together Community section!

Meme of the week!

This is literally it 😂. Meme shared by Robino > Toi#0475.

Featured Community post from the Discord

Spidey#2169 built an app so that anyone can extend GPT-3 with new skills/capabilities and he’s looking for feedback. Check it out here and support a fellow member of the community! You can leave your feedback in the thread here.

Read the post on Discord.

AI poll of the week!

Join the discussion on Discord.

Towards AI Article of the week

A Quick Setup for Neural Networks Hyperparameters for Best Results by Youssef Hosni

It is not easy to find optimal hyperparameters settings and test different settings by ourselves until achieving good results in neural networks. In general, we tend to spend a lot of time tweaking these hyperparameters settings and getting better models. However, the most important thing is to avoid overfitting. Here is a quick guide on how to set up hyperparameters in a short time to get good results.

Other must-read articles

Evaluating Mode Collapse in GANs Using NDB Score by Shashank Kumar

How To Compare 2 Datasets With Pandas-profiling by Fabiana Clemente

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.

Ethical Take on Meta’s CICERO by Lauren

Meta has announced CICERO, the first AI to achieve human-level performance in the popular strategy game Diplomacy. Using NLP, CICERO negotiates, persuades, and collaborates with human players to achieve strategic goals. This advancement has huge implications for the future of how we interact with AI. Even the best conversation bots reach a breaking point or spit out an incoherent line when prompted (or not prompted). CICERO is not perfect and is subject to these flaws as well, but its greater skills of negotiation, persuasion and collaboration add a layer of conversational complexity to our AI interactions that are entirely unprecedented. This might even look like reasoning in its limited machine context.

What really makes CICERO unique (and concerning) is the junction of its human behavior prediction abilities and its linguistic persuasion, a recipe for groundbreaking conflict resolution or unprecedented and direct manipulation when scaled. Its open-sourcing is a comfort to know that a level of transparency is being upheld, but the context CICERO is applied in will determine where it falls on this spectrum. I would like to see some guidelines in place for the use cases of CICERO, as Meta’s mere hope that the research will be built on responsibly by the community is devoid of responsibility for what could come of its future uses. Strategic goals can easily become nefarious ones, but I too hope that good will come of this invention.

Job offers

Our sponsored job offer this week comes from NannyML who are looking for Data Science Writer Interns to work on content closely with senior Data Professionals and the founding team:

Data Science Writer Intern @ NannyML (Remote)

Senior Data Scientist / AI Developer @ Uniphore (Spain, Hybrid Remote)

Lead Data Engineer, Data Platform @ Tubi (Remote)

Interested in sharing a job opportunity here? Contact [email protected].

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


This AI newsletter is all you need #23 was originally published in Towards AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

Join thousands of data leaders on the AI newsletter. It’s free, we don’t spam, and we never share your email address. Keep up to date with the latest work in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI

Feedback ↓