AI innovations that made headlines in 2021

0



In the meantime, AI has proven its efficiency and effectiveness. The field of artificial intelligence is constantly evolving and improving every day. Tech companies and researchers invest heavily in innovation as the impact of AI on the world’s biggest problems has enormous potential. In late 2021, let’s look back at some of the top AI innovations and incidents that took center stage this year.

January

DALL · E. by OpenAI

OpenAI released DALL · E, a 12 billion parameter version of GPT-3 trained to generate images from text descriptions using a data set of text-image pairs. OpenAI said that DALL · E is a Transformer language model that receives both the text and the image as a single data stream of up to 1280 tokens. It added that DALL.E can render an image from scratch and also modify aspects of an image using text prompts.

Register for this Session>>

For more information, see here.

February

TensorFlow 3D

Google published TensorFlow 3D (TF 3D) designed to integrate 3D deep learning capabilities into TensorFlow. The tech giant said TF 3D is packed with popular operations, loss functions, and data processing functions that help develop, train, and deliver 3D models for understanding scenes.

For more information, see here.

March

Metas SEER

Meta-AI released SEHER (SElf-supERvised), a self-monitored computer vision model with billions of parameters that can learn from any group of images on the Internet. It does not require the careful curation and labeling that most computer vision training models do. The company also said that SEER also outperformed cutting-edge monitored models in the downstream such as low-shot, object detection, segmentation and image classification.

For more information, see here.

April

EU draft regulation for AI

The EU executive EU Commission has published a proposal for a regulation on artificial intelligence. Among other things, this is intended to define mechanisms and restrictions for the use of AI, its violations, and regulatory requirements for AI.

For more information, see here.

Can

Google’s Vertex AI

In May, Google announced the general availability of at the Google I / O event. known Vertex AI. It is a managed machine learning platform that will enable companies to accelerate the deployment and maintenance of AI models, Google claimed. Google also said that Vertex AI needed almost 80% fewer lines of code to train a model compared to competing platforms. It will help data scientists and ML engineers implement machine learning to create and manage ML projects throughout the development cycle.

For more information, see here.

June

GitHub copilot

OpenAI and Microsoft’s GitHub Copilot is an AI pair programmer to write better code. GitHub Copilot works with different languages ​​like Python, JavaScript, TypeScript, Ruby, Java and Go. GitHub Copilot can be used as an extension on the desktop or in the cloud on GitHub Codespaces. The company said the programmer can use the copilot to review alternative suggestions, choose what to accept and reject, and manually edit suggested code.

For more information, see here.

July

DeepMind Open Source AlphaFold 2.0

The AlphaFold 2.0 code was created by DeepMind as open source. This AI algorithm predicts the shape of proteins, which is a major challenge in life sciences. In 2018, AlphaFold 1.0 was released, although it turned out to be not good enough to keep researchers in the field busy. After further improvements, AlphaFold 2.0 was released in December 2020 and has received a lot of recognition. By publishing the source code, DeepMind aims to provide the scientific community with better research opportunities in areas such as drug discovery.

More information can be found here.

August

Tesla AI day

At Tesla AI Day, Elon Musk announced that the company was working on a humanoid robot. He added that Tesla will build a robot in human form that could perform repetitive tasks, and the prototype will likely be ready by next year. The code name for the bot is ‘Optimus’. Tesla’s director Ganesh Venkataramanan showed the computer chip Tesla uses to run its supercomputer dojo. It contains 7 nm technology and is equipped with 362 teraflops of computing power.

See also
Top 10 Research Papers on Federated Learning

More information can be found here.

September

Toshiba’s VQA AI

Toshiba Corporation introduced the Visual Question Answering (VQA) AI, which can recognize people and objects, colors, shapes, looks and background details in images. This mechanism solves the problem of answering questions about the positioning and appearance of people and objects. It can acquire the information necessary to handle a wide variety of questions and answers, and can find uses for a wide variety of uses without customization requirements.

For more information, see here.

October

Facebook is renamed Meta

In a big step, CEO Mark Zuckerberg announced that Facebook had changed its name to Meta at Connect 2021, which took place recently. Meta’s reach will extend well beyond social media. The metaverse gives the feeling of a hybrid structure of online social experiences that are extended to the physical world.

For more information, see here.

November

NVIDIA Omniverse

NVIDIA has jumped on the Metaverse bandwagon too. It announced Omniverse VR, where developers, designers, researchers and engineers can combine key design tools and assets to work together in a common virtual space. Omniverse Avatar, a new platform for creating interactive AI avatars using computer vision, NLP and simulation technologies, was also introduced. The company also showed us the NVIDIA Omniverse Replicator, a synthetic data generation engine used to train deep neural networks. NVIDIA also announced that Omniverse Enterprise is now generally available.

For more information, see here.

Isomorphic Labs started

Alphabet announced the launch of Isomorphic Labs, which aims to accelerate the drug discovery process. Demis Hassabis is the founder and CEO of Isomorphic Labs. He posted on Twitter that the goal is to re-engineer the drug discovery process from the first principles with an AI-first approach.
For more information, see here.


Subscribe to our newsletter

Get the latest updates and relevant offers by sharing your email.


Join our Telegram Group. Be part of an engaging community


Share.

Comments are closed.