Tutorials Day
February 3
(separate registration, online format)
NLP & CV Day
February 4
RL & AGI Day
February 5
Tutorials Day
(separate registration, online format)
February 3
Pavel Fakanov, Replika
Mikhail Guschin, HSE
Vadim Strijov, MIPT
Kirill Aksenov, MIPT
Dmitry Mironov, NVIDIA
Alexey Gruzdev, Intel
Olga Perepelkina, Intel
Dmitry Soshnikov, Microsoft
Ales Kuchumol, Huawei
NLP & CV Day
Thursday, February 4
09:00 – 10:00
09:00 – 10:00
Registration
10:00 – 10:10
10:00 – 10:10
Conference and day opening
Igor Pivovarov, OpenTalks.AI
What will be at the conference, main ideas, numbers, highlights.
10:10 – 11:40
10:10 – 11:40
Computer vision - overview of 2020 results
Big conference hall
Computer vision - overview of 2020 results
Alexey Dosovitsky, Google Brain
Vision in 2020: Transformers, Transfer, and Beyond
3D scenes modeling: new approaches in 2020
Victor Lempitsky, Samsung, Skoltech
In 2020, the computer vision community actively explored new approaches to modeling 3D scenes and objects based on visual data. As part of the report, we will look at these approaches, their main features and differences.
11:40 – 12:00
11:40 – 12:00
Coffee-break
12:00 – 13:00
12:00 – 13:00
Parallel sessions
NLP R&D
ML - science

Dmitry Vetrov, Samsung AI, HSE
Loss Fractality, Double Descent Effect, and Power Laws in Deep Learning - Pieces of the Same Mosaic
Sergey Nikolenko
POMI RAS,
Neuromation
Understanding the Multimodal World: Recent Results, Tasks and Challenges
Denis Timonin,
NVIDIA
Megatron-LM: Teaching Multi-Billion LMs Using Model Parallelism Technique
CV in healthcare
Features of AI in medical diagnostics: data, validation and practical application
Kirill Arzamasov,
NPKTs DiT DZM
Neural network algorithms to improve the quality of medical images
Alexey Chernyavsky, Philips
Artificial intelligence and modern challenges: the most interesting solutions for CT diagnostics of COVID-19 through the prism of global technological trends
Inna Moroz,
Care Mentor AI
Alexander Gromov,
Third Opinion
Stability of neural network models in the analysis of CT / LDCT studies
Egor Filimonov,
Huawei
Huawei Atlas platform capabilities and efficient heterogeneous inference
Valentin Malykh, Huawei
Moderator
Reference architecture of a service center robot in industries with fluid business processes
Alexander Prozorov, Cybersyn
Sergey Morozov,
NPCMR
Moderator
Sergei Kuznetsov,
HSE
Moderator
Natalia Lukashevich, MSU, BMSTU
Sentiment Analysis in Relation to the Company - What BERT Failed
13:00 – 13:15
13:00 – 13:15
Coffee-break
13:15 – 14:00
13:15 – 14:00
Parallel sessions
NLP R&D
ML - scientific reports

Andrei Ustyuzhanin, HSE
The frontier of simulation-based inference
Predictive analytics in business
"CAPTAIN" of the Digital Arctic
Sergey Orlov,
Gazprom Neft
Application of machine learning on the example of flotation
Vadim Sidelnikov, Nornikel
Andrei Kotseruba,
VTB
Semi-supervised Geo2Vec
Sequential analysis in image recognition: how to improve speed of inference and classification
Alexey Zorchenlov,
Huawei
Using NLP Methods to Work with Program Code
Emeli Dral, Evidently.AI
Moderator
Andrei Savchenko,
HSE
Artem Artemov,
Cognitive systems
B2NLP framework.
Model for determining the emotional tone of the text
Valentin Malykh, Huawei
Moderator
Fake news and other types of potentially dangerous discourse: typology, approaches, datasets, competitions
Konstantin Vorontsov, MIPT
Sergei Kuznetsov,
HSE
Moderator
14:00 – 15:15
14:00 – 15:15
Lunch
We have maximized the lunch time to 1 hour and 15 minutes to reduce queues of offline participants.
15:15 – 16:30
15:15 – 16:30
NLP- overview of 2020 results
Big conference hall
Top NLP Results in 2020
Grigory Sapunov, Intento
An overview of the most important works and results in the field of natural language processing in 2020.
How AI impacted business in 2020
Sergey Lukashkin, VTB
This year we will not review how CV or NLP are applied in business. Instead, we'll take a look at how global AI trends have emerged in 2020 and how they will impact businesses this year and the next one.
16:30 – 16:50
16:30 – 16:50
Coffee-break
16:50 – 17:50
16:50 – 17:50
Parallel sessions
CV R&D
ML - science

NLP in business
Life-giving analytics. DODO contact center automation.
Optimization of neural networks and their development
Ilya Zharikov,
MIPT
Speed up of convolutional networks using quantization. Quantization aware training
Dmitry Pagin, TrafficData
Analysis of the efficiency of pattern recognition on non-standard types of images using the example of radar images and X-ray images of baggage and hand luggage
Nikita Andriyanov, Financial University
Roman Doronin, EORA.AI
Optimization of business processes and workflow by using NLP technologies
Business case: Digital Auditor

Maxim Milkov,
Softline
Associative inference tensor machine
Sergey Terekhov,
RAII, RANI
Influence of AI technologies on the development of machine-readable document flow in Russia
Search for interpretable semantics in various types of signals. Applications to real products.
Denis Parhomenko, Huawei
Anna Serebryanikova, nlogic
Moderator
Arkady Sandler,
MTS
Moderator
Alexander Krainov,
Yandex
Yuri Vizilter,
GOSNIIAS
Presentation of Plat, a russian machine learning ecosystem
17:50 – 18:05
17:50 – 18:05
Coffee-break
18:05 – 19:05
18:05 – 19:05
Parallel sessions
CV R&D

AGI - science

Predictive analytics in healthcare
GeoSense: AI Helps Track Health and Aging Using Your Smartphone
Measuring progress towards AGI with tests
Dmitry Salikhov, Sber

Sergey Schevchuk, Third Opinion
Models for predicting the occurrence of complications and risks of severe course of Covid-19 based on data from case histories and X-ray studies of patients
Timofey Pyrkov,
Gero
VisionLabs ATAC - multicam people tracking
Daniil Kireev,
VisionLabs
Alexander Gusev,
Webiomed
Prospects for predictive analytics in reducing cardiovascular morbidity and mortality
AutoDL or how to reduce development and production costs of neural networks
Sergey Alyamkin,
Expasoft
Is it possible to train an accurate medical AI model on incomplete data of questionable quality?
Evgeny Nikitin,
Celsus
Olga Perepelkina, Intel
Federated Learning for Healthcare: A privacy-preserving method to train data science models on sensitive datasets
Moderator
Boris Zingerman, INVITRO
Moderator
Vitaly L. Dunin-Barkovsky, NIISI RAS

Hierarchical Sequence Memory
Oleg Serebrennikov
Logical decision-making systems "mivar brains" for intelligent autonomous robotic systems, cyber-physical systems and the Internet of Things.
Oleg Varlamov,
BMSTU, Mivar
Moderator
Alexander Krainov, Yandex
CNN Robustness research: Application to face detectors and face ID systems
Aleksandr Petiushko,
Huawei
RL & AGI Day
Friday, February 5
09:00 – 10:00
09:00 – 10:00
Registration
10:00 – 10:10
10:00 – 10:10
Opening day. AI in Russia - some trends
Igor Pivovarov, OpenTalks.AI
An independent conference is a mirror of the industry, reflecting its main trends and patterns. What are these trends and what can be said about the current state of AI in Russia - about this at the opening of the third day.
10:10 – 11:40
10:10 – 11:40
Reinforcement Learning - main results of 2020
Big conference hall
RL in robotics
Alexander Panov, Moscow Institute of Physics and Technology, Institute of Applied Mathematics and Research
An overview of the Reinforcement Learning applications in robot navigation - basic methods, key competitions, results.
Review of main works and results in RL in 2020
Alexander Novikov, DeepMind
Overview of top results in Reinforcement Learning in 2020
11:40 – 12:00
11:40 – 12:00
Coffee-break
12:00 – 13:00
12:00 – 13:00
Parallel sessions
Data markup for ML
Predictive analytics in business
Olga Megorskaya,
Yandex
We are responsible for what we've learned: data markup as a key part of building AI products
Valentin Biryukov, Yandex
ML in preparing data for ML: predicting the quality of the work of performers when marking up data
CV - science

Anton Konushin,
HSE, Samsung
Depth Map Assessment
Evgeny Burnaev,
Skoltech
Deep Vectorization of Technical Drawings
Image markup for Japanese startup
Roman Kutsev, Trainingdata.ru
Andrey Arefiev, InfoWatch
Machine learning use cases for information security
Alexey Tsyplakov, IPG
AI for expert systems in aviation and beyond
Extraction and analysis of price factors for monitoring the grain market
Alexznder Khaidarov, MITLabs
How tolokers help to improve product name translation in AliExpress Russia
Andrei Olkhovik, AliExpress Russia
Hybrid intelligence for data markup tasks
Ivan Bolokhov, Cognitive Systems
Maxim Fedorov,
Skoltech
Moderator
Alexey Dral,
BigData Team
Moderator
Application of deep learning methods to automatical detection of defects in trunk pipelines
Adele Yarullin, Innopolis University
Aircraft engine blade defects recognition
Maxim Rassabin, Innopolis University
Daria Baidakova,
Yandex
Moderator
13:00 – 13:15
13:00 – 13:15
Coffee-break
13:15 – 14:00
13:15 – 14:00
Parallel sessions
Hardware for
Deep Learning

NLP R&D for legal practice
Grigory Sapunov, Intento
AI hardware overview
Yulia Schevchenko,
MITLABS
Themis and transformers: the use of linguistic models based on the "transformer" architecture in legal practice.
Ignat Postny,
TAG Consulting
Why is machine learning hampering the development of Legal AI?
Научные доклады - AGI
Anton Kolonin,
Aigents

Justification of the neurosymbolic architecture of general artificial intelligence using the example of reinforcement learning
Evgeny Vityaev,
IM SB RAS
A problem-solving approach to general artificial intelligence

Moderator
Vladimir Smolin, Keldysh Institute of Applied Mathematics
Moderator
Andrei Neznamov, Robopravo
Elena Tutubalina,
KFU
RuREBus-2020 Shared Task: Russian Relation Extraction for Business
14:00 – 15:15
14:00 – 15:15
Lunch
We have maximized the lunch time to 1 hour and 15 minutes to reduce queues of offline participants.
15:15 – 15:45
15:15 – 15:45
Robofootball
A real football match between the robots of the Starkit team (MIPT), championship winners:
  • 1st place at Robocup Asia-Pacific 2019 (Humanoid Kid Size)
  • 3rd place at FIRA 2019–3rd (Sprintand Obstaclerun)
  • 1st place at Robocup 2019 (SPL Challenge Shield)
15:45 – 16:00
15:45 – 16:00
Coffee-break
16:00 – 17:00
16:00 – 17:00
Parallel sessions
Rec. systems and predictive analytics - R&D

Investments and startups in AI
Mikhail Guschin,
HSE
Detecting temporary changes for predictive analytics systems
Artem Prosvetov,
LANIT / CleverDATA

User vectorization and our steps towards Federated Learning
Autonomous systems - science
Alexey Okunev,
NSU
Reconstruction of geographic coordinates of recognized objects based on SLAM and Deep Learning technologies during survey video from UAVs
Vladislav Kibalov,
Evocargo
Safe Speed Control and Collision Probability Estimation Under Ego-Pose Uncertainty for Autonomous Vehicle
Evgeny Tsymbalov, Huawei
Uncertainty assessment and decision making: metrics, methods and applicability
Sergey Sviridov,
Optimate AI
Moderator
Sergey Khodakov,
Skolkovo
Tech Explorer program - helping companies develop
AI
Evgeny Sheenko,
Skolkovo
Investments in AI companies. Priority areas and success stories
SeaScan AI technology for identifying underwater objects
Maria Mashkeeva,
Electronic platforms
Examus: AI-based proctoring system
Pavel Krivozubov,
Skolkovo
Moderator
Sergey Zakharov,
Cera marketing
Cera and the smart supermarket
Roman Fedorenko,
Innopolis University
What tasks do autonomous drones have and how we solved them. The example of the Airbot competition
insolver.io - open-source low-code framework for insurance calculations automation
Frank Shikhaliev, Mindset
Rinat Sadekov,
MISIS
Moderator
Ilya Pronin, Scan Graf
17:00 – 17:15
17:00 – 17:15
Coffee-break
17:15 – 19:00
17:15 – 19:00
Strong artificial intelligence - AGI
Big conference hall
It is difficult to speak about the results in the field of AGI, since there are no working AGI models. We will talk about the current state of research at AGI, look at the advantages and disadvantages of existing approaches, and propose new models. This session will open up new paths for you to build AGI!
AGI architectures and approaches - where we are now and where to go next
Igor Pivovarov, CAIST MIPT, OpenTalks.AI
Review of the main architectures and approaches for building AGI, biological and mathematical aspects, neural network and cognitive architectures. Generalization of recent results and new models.
Reverse engineering of the brain as a path to strong AI
Sergey Shumsky, Russian Association of Neuroinformatics
We will consider how the brain works from the point of view of machine learning, which blocks are responsible for what and how they are modeled. Let's outline the paths to new models.
Neurorealistic Artificial Intelligence: The Missing Links
Konstantin Anokhin, Moscow State University
Artificial intelligence is driven by the idea of replicating the fundamental principles of natural intelligence. We will consider: a) key features of biological intelligence, which so far elude reproduction in artificial systems, b) already known principles of structure and functioning of natural neural networks that have not yet entered artificial neural networks. This knowledge could help us to create the next generation of artificial intelligence.
19:00 – 22:00
19:00 – 22:00
Afterparty
You will have a wonderful opportunity to communicate informally with the speakers and participants of the conference and listen to the performances of the music groups of the AI industry companies!