Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
OpenTalks.AI - Сергей Шумский, Символьное мышление роботов
Search
OpenTalks.AI
February 14, 2019
Science
0
620
OpenTalks.AI - Сергей Шумский, Символьное мышление роботов
OpenTalks.AI
February 14, 2019
Tweet
Share
More Decks by OpenTalks.AI
See All by OpenTalks.AI
OpenTalks.AI - Виктор Лемпицкий, Моделирование 3Д сцен: новые подходы в 2020 году
opentalks
0
420
OpenTalks.AI - Алексей Чернявский, Нейросетевые алгоритмы для повышения качества медицинских изображений
opentalks
0
400
OpenTalks.AI - Александр Громов, Устойчивость нейросетевых моделей при анализе КТ/НДКТ-исследований
opentalks
0
350
OpenTalks.AI - Денис Тимонин, Megatron-LM: Обучение мультимиллиардных LMs при помощи техники Model Parallelism
opentalks
0
460
OpenTalks.AI - Егор Филимонов, Возможности платформы Huawei Atlas и эффективный гетерогенный инференс.
opentalks
0
110
OpenTalks.AI - Александр Прозоров, Референсная архитектура робота сервисного центра в отраслях с изменчивыми бизнес-процессами
opentalks
0
340
OpenTalks.AI - Наталья Лукашевич, Анализ тональности по отношению к компании — с чем не справился BERT
opentalks
0
320
OpenTalks.AI - Константин Воронцов, Фейковые новости и другие типы потенциально опасного дискурса: типология, подходы, датасеты, соревнования
opentalks
0
390
OpenTalks.AI - Дмитрий Ветров, Фрактальность функции потерь, эффект двойного спуска и степенные законы в глубинном обучении - фрагменты одной мозаики
opentalks
0
410
Other Decks in Science
See All in Science
Ph.D. defense "Convex Manifold Approximation for Tensors"
gkazunii
0
180
Direct Preference Optimization
zchenry
0
150
AI科学の何が“哲学”の問題になるのか ~問いマッピングの試み~
rmaruy
1
1.3k
ABEMAの効果検証事例〜効果の異質性を考える〜
s1ok69oo
3
1.5k
AI(人工知能)の過去・現在・未来 —AIは人間を超えるのか—
tagtag
0
230
DEIM2024 チュートリアル ~AWSで生成AIのRAGを使ったチャットボットを作ってみよう~
yamahiro
3
660
JSol'Ex : solar image processing in Java
melix
0
250
Machine Learning for Materials (Lecture 4)
aronwalsh
0
680
データで課題を解決する -因果関係を調べる統計的因果推論-
sshimizu2006
4
1.3k
Demucsを用いた音源分離
508shuto
0
190
Machine Learning for Materials (Lecture 2)
aronwalsh
0
590
SIGDIAL論文読み会: PGTask: Introducing the Task of Profile Generation from Dialogues
kaiyo3
0
100
Featured
See All Featured
Large-scale JavaScript Application Architecture
addyosmani
504
110k
BBQ
matthewcrist
80
8.8k
GraphQLとの向き合い方2022年版
quramy
33
12k
Creating an realtime collaboration tool: Agile Flush - .NET Oxford
marcduiker
14
1.5k
Ruby is Unlike a Banana
tanoku
96
10k
XXLCSS - How to scale CSS and keep your sanity
sugarenia
242
1.2M
The Art of Programming - Codeland 2020
erikaheidi
43
12k
Side Projects
sachag
451
41k
Building Applications with DynamoDB
mza
88
5.6k
The MySQL Ecosystem @ GitHub 2015
samlambert
244
12k
GitHub's CSS Performance
jonrohan
1025
450k
Become a Pro
speakerdeck
PRO
12
4.6k
Transcript
The Symbolic Mind of Robots Sergey Shumsky, 2019
Motivation: common architecture of robotic brains Reinforcement learning
Motivation: common architecture of robotic brains
Agenda I. Problem Deep Reinforcement Learning II. Solution Deep Symbolic
Learning III. Prospects Robot Operating System
Agenda I. Problem Deep Reinforcement Learning
AlphaZero: intuition + calculation ▪ Deep intuition: Yes ▪ Deep
planning: No ▪ Complexity of learning ∝ 2 , 3 TFLOPSyears – weights update 500 TFLOPSyears – search options
Google DeepMind research program "One way you can think about
our research program is: 'Can we build out from our perception, using deep-learning systems and learning from first principles? Can we build out all the way to high-level thinking and symbolic thinking?' " D. Hassabis (Google DeepMind)
Agenda I. Problem Deep Reinforcement Learning II. Solution Deep Symbolic
Learning
Deep symbolic learning ▪ Hierarchy of plans state sequences ▪
Planning and Learning in real time ▪ Complexity of learning ∝ ∝
Images and Symbols Coding Decoding Image ~ 106 bits Sensors
States Actions Effectors Image ~ 106 bits Symbols ~ 5 bits Planning
The trick: symbolic coding of images 2106 ~ , →
1 2 … > ( = 30, k = 7) symbols images = 210 <
Symbolic thinking ~ planning ▪ Coding of sequences ~ k
variants (words) ~ 30 Symbolic sequences can be remembered symbol +1 +2 ~ 30 ~ 30
Symbolic thinking ~ planning символ +1 +2 Image code Sequence
code (pattern)
Deep symbolic learning Layer L Layer L -1 1 2
…
Deep symbolic learning ▪ Learning useful patterns of interaction with
the world ▪ At all hierarchical levels ▪ In real time Encoder- Decoder +1 +1 Parser +1 Encoder- Decoder Parser symbols patterns symbols
Agenda I. Problem Deep Reinforcement Learning II. Solution Deep Symbolic
Learning III. Prospects Robot Operating System
Prospects Sensory intelligence Strategic intelligence Robotic intelligence Goal
setting and planning to achieve them Achieving the goal
Open AI gym Mountain car (Igor Pivovarov)
Interested? Join us!
[email protected]
[email protected]