XLS-R: FAIR speech recognition model in 128 languages

21 November 2021

XLS-R: FAIR speech recognition model in 128 languages

FAIR has posted the code of XLS-R, a self-learning model that performs speech recognition, to the public. XLS-R supports 128 languages and surpasses all previous multilingual models in benchmarks. The…

MetNet-2: Google’s 12-hour weather prediction model

16 November 2021

MetNet-2: Google’s 12-hour weather prediction model

Google AI has introduced MetNet-2 — an improved version of the MetNet weather prediction model. MetNet-2 allows you to forecast the weather for 12 hours ahead with a spatial resolution…

PARP: algorithm for pruning speech recognition models

6 November 2021

PARP: algorithm for pruning speech recognition models

PARP is the most effective method among analogues for cropping large speech recognition models. PARP can be used for automatic speech recognition in rare languages. Speech recognition technology has become…

Language models mimic the work of the brain

29 October 2021

Language models mimic the work of the brain

Scientists have revealed the similarity of the work of natural language processing models with brain activity during the perception of language information. In particular, predicting the next word based on…

The Google SimVLM model generates image captions

25 October 2021

The Google SimVLM model generates image captions

Google has introduced a SimVLM model that generates text from a single image. SimVLM features include forming a simple sentence description, completing a sentence with the first few words, and…

The neural network was trained to simulate the olfactory system of a fruit fly

20 October 2021

The neural network was trained to simulate the olfactory system of a fruit fly

Scientists have created a three-layer neural network that classifies odors. After training, the model of communication between neurons accurately reproduced the structure of the olfactory system of a fruit fly.…

Neural network trained to generate makeup to bypass facial recognition systems

4 October 2021

Neural network trained to generate makeup to bypass facial recognition systems

Israeli researchers have presented a model that generates makeup to bypass facial recognition systems. After applying cosmetics in accordance with the patterns presented by the neural network, the faces of…

The Open AI model generates short retellings of books

28 September 2021

The Open AI model generates short retellings of books

OpenAI has trained a neural network to retell the contents of books. The model is based on GPT-3 and uses the method of recursive decomposition, that is, first creates retellings…

IC-GAN: generating synthetic images without training individual objects

21 September 2021

IC-GAN: generating synthetic images without training individual objects

IC-GAN is a set of FAIR models for generating images, objects and scenes on which were absent in the training dataset. IC-GAN can be used for data augmentation. Generative-adversarial networks…

Invisible keyboard for smartphones based on decoders

18 September 2021

Invisible keyboard for smartphones based on decoders

Scientists have developed a model that allows you to enter text on smartphones based on the user’s memory of the location of letters in the keyboard layout. This keyboard allows…

The neural network was trained to simulate a biological neuron

12 September 2021

The neural network was trained to simulate a biological neuron

Scientists have chosen the architecture of a neural network that reproduces the complexity of a biological neuron with 99% accuracy. The study allowed for the first time to compare biological…

Google AI model recognizes diseases of the chest organs

5 September 2021

Google AI model recognizes diseases of the chest organs

The deep learning algorithm developed by Google AI recognizes a wide class of diseases from chest X-rays. The model was able to recognize abnormal images of patients with diseases that…

MIT has trained drones to fly around obstacles at high speeds

30 August 2021

MIT has trained drones to fly around obstacles at high speeds

MIT engineers have solved the problem of flying drones over obstacles at high speeds. The algorithm allows you to quickly perform operations such as search and rescue of people after…

Neural network has discovered new peptides for drug development

23 August 2021

Neural network has discovered new peptides for drug development

Researchers at the Massachusetts Institute of Technology used a convolutional neural network to search for highly active and non-toxic peptides. The neural network proposed previously unknown peptides that allowed to…

MIT has developed a probabilistic system for evaluating AI fairness

16 August 2021

MIT has developed a probabilistic system for evaluating AI fairness

SPPL is an open-source probabilistic programming language developed at MIT and aimed at accurately assessing the degree of artificial intelligence bias against certain groups of people. SPPL is implemented in…

LiftPose3D: Evaluation of a 3D pose from a single image

9 August 2021

LiftPose3D: Evaluation of a 3D pose from a single image

A group of scientists from the Federal Polytechnic School of Lausanne presented LiftPose3D – a neural network that restores a 3D pose from an image from one angle. The model…

Droidlet: FAIR platform for the development of intelligent agents

1 August 2021

Droidlet: FAIR platform for the development of intelligent agents

FAIR presented Droidlet , an open-source platform aimed at simplifying the integration of a wide range of modern machine learning algorithms with the field of robotics. Droidlet allows you to…

The NASA neural network calibrates images from space telescopes

28 July 2021

The NASA neural network calibrates images from space telescopes

Space telescopes require frequent recalibration due to their degradation associated with intense radiation exposure and particle bombardment. The NASA neural network allows us to solve this problem remotely, opening up…

The neural network was trained to predict cell growth

23 July 2021

The neural network was trained to predict cell growth

Scientists from the Institute of Industrial Sciences at the University of Tokyo have developed a machine learning algorithm that allows you to predict the size of an individual cell as…

BlenderBot 2.0: FAIR chatbot with long-term memory

18 July 2021

BlenderBot 2.0: FAIR chatbot with long-term memory

FAIR has introduced the second version of BlenderBot , an open-source chatbot that combines several conversational skills, including personification, empathy and factual knowledge. BlenderBot 2.0 has a long-term memory and…

Artificial intelligence allowed to study the boiling process

11 July 2021

Artificial intelligence allowed to study the boiling process

MIT researchers have trained a neural network to predict the boiling crisis of liquids from images from infrared cameras. Potential applications of the technology include automation of the cooling process…