Inside Artificialis - #17
Jun 30, 2023 10:14 am
Hey, . Welcome to the monthly version of the Artificialis newsletter!
A summary of what happened during the month, the best blogs of our Medium publication, latest news of our Discord server events, and recent news in the world of Artificial Intelligence and Machine Learning!
So sit back, take a cup of coffee, let's go over what happened this month.
A little bit from me and the server:
I started being a bit more serious on Twitter, trying to post daily. It'll be a mix of what you see on Discord, LinkedIn, and some short-form content. Make sure not to miss it!
Some new blogs:
- The Future is Self-Supervised: An Introduction to DINOv2, by Alessandro
- Getting Started with Depth Estimation using MiDaS and Python, by Nabeel Khan
AI in the world
EU lawmakers pass landmark artificial intelligence regulation
The European Union’s AI Act is the first comprehensive set of regulations for the artificial intelligence industry.
The law proposes requiring generative AI systems, such as ChatGPT, to be reviewed before commercial release. It also seeks to ban real-time facial recognition.
It comes as global regulators are racing to get a handle on the technology and limit some of the risks to society, including job security and political integrity.
I-JEPA: The first AI model based on Yann LeCun’s vision for more human-like AI
This model, the Image Joint Embedding Predictive Architecture (I-JEPA), learns by creating an internal model of the outside world, which compares abstract representations of images (rather than comparing the pixels themselves). I-JEPA delivers strong performance on multiple computer vision tasks, and it’s much more computationally efficient than other widely used computer vision models. The representations learned by I-JEPA can also be used for many different applications without needing extensive fine tuning.
Sir Paul McCartney says artificial intelligence has enabled a 'final' Beatles song
He told BBC Radio 4's Today programme the technology had been used to "extricate" John Lennon's voice from an old demo so he could complete the song.
Robots learn to perform chores by watching YouTube
Researchers at Carnegie Mellon University (CMU) are using video footage from YouTube to teach robots how to perform tasks. The system uses video of humans executing tasks, and the robot learns and adapts to replicate the actions. By using online videos, robots can access a vast amount of training data to improve their ability to navigate and operate in unpredictable environments.
Introducing 22 system cards that explain how AI powers experiences on Facebook and Instagram
As part of Meta’s commitment to transparency, today we are sharing 22 system cards that contain information and actionable insights everyone can use to understand and customize their specific AI-powered experiences in our products. We are releasing these cards to help people better understand AI’s role in many Instagram and Facebook features, and to explain how people’s choices and behaviors influence what content they see through our ranking and recommendation systems, such as a new video or a creator they may want to follow.
Highlights from CVPR - 2023
CVPR stands for the Conference on Computer Vision and Pattern Recognition. It is an annual conference that presents the latest research on computer vision, pattern recognition, image processing, and machine learning. This event attracts scholars, researchers, and professionals who are interested in these fields. It is considered one of the most important events in the field, with a highly competitive peer-review process. Papers accepted to CVPR often represent significant advancements in these domains. Here's some highlights:
- OneFormer: One Transformer To Rule Universal Image Segmentation
- X-Decoder: Generalized Decoding for Pixel, Image and Language
- Vid2Avatar: 3D Avatar Reconstruction from Videos in the Wild via Self-supervised Scene Decomposition
- 3D-aware Conditional Image Synthesis
- DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation
- High-resolution image reconstruction with latent diffusion models from human brain activity
- Improving Visual Representation Learning through Perceptual Understanding
Tip of the month
Approximately 330 million terabytes of data are created each day.
Videos account for over half of internet data traffic.
Can you imagine keeping up in datasets curation if we stick to supervised learning? I don't.
I believe the future of computer vision is self-supervised.
↓
Check out the rest of the thread
— Alessandro Lamberti (@alessandro__ml) June 27, 2023
'Till next month, you can find everyone here:
Have a fantastic month, !
If you'd like to support our community, and have access to millions of amazing articles and tutorials, consider subscribing to Medium's program via our link, we'll receive a small portion of your fee.
All of the money will be used to sponsor our Discord's events prizes!
