AIRI TeaminAIRI.InstituteUsing the general cost functional made the optimal transport with class preservation more accurateIn the mid-20th century, Soviet mathematician Leonid Kantorovich developed ideas for optimal solutions in economics, including the problem…Jul 29Jul 29
AIRI TeaminAIRI.InstituteThe new framework required 50 times less data to predict the optimal geometry of molecules using…To create a new drug or material, scientists must research molecules that have yet to be synthesized and hope to find valuable properties…Jul 11Jul 11
AIRI TeaminAIRI.InstituteLanguage models for chemical tasks turned out to be vulnerable to changes in molecule…To predict the outcome of a chemical reaction, one can describe it as the interaction of a multitude of neutral atoms, ions, and electrons…Jul 2Jul 2
AIRI TeaminAIRI.InstituteGaussian mixtures helped to improve the learning of object-centric scene representationThe use of neural networks has led to significant progress in the field of computer vision and other areas where segmenting objects in…Jun 17Jun 17
AIRI TeaminAIRI.InstituteEnergy-based models made Schrödinger Bridge lighterSince the early 2000s, energy-based models (EBMs) have been gaining popularity in machine learning. They are based on the application of…May 15May 15
AIRI TeaminAIRI.InstituteRecurrent memory removed restrictions on the length of input sequences for transformersToday, natural language processing is unthinkable without the use of transformer models. Unlike recurrent neural networks, which solved…Apr 19Apr 19
AIRI TeaminAIRI.InstituteBulk Search Improved Anonymous Multi-Agent Path FindingIt is difficult to argue with the thesis that soon, many everyday tasks will be performed by robots controlled by various artificial…Apr 11Apr 11
AIRI TeaminAIRI.InstituteInformation about nearest neighbors helped complete knowledge graphs more accuratelyKnowledge graphs are a way of organizing structured information, in which entities are represented as nodes of a multi-relational graph…Apr 2Apr 2
AIRI TeaminAIRI.InstituteComplex dynamics of anisotropy and intrinsic dimensions have been discovered in transformersTransformer-based models (transformers), proposed in 2017 by Google Brain researchers, revolutionized both natural language processing and…Mar 28Mar 28
AIRI TeaminAIRI.InstituteTwo new approaches accelerated Multi-agent Pathfinding in decentralized settingsThe task of Multi-agent Pathfinding (MAPF) arises whenever a group of agents needs to move through a graph to achieve their goals without…Mar 15Mar 15