Bismillah

---
open ai


go game school
open ai
dota players
---
chess
dota player
---

alpha mind google
google deep mind


gui zihao
---
general purpose functionsl
---
great ability to learn
---
augment our learn
---
huble telescope
---
alpha go
---

general purpose learning system
---
health care
robotic
optimise data centre
---
go game
basic tutorial

19 x 19 professional

13 x 13

9 x 9
--
2 player game
take turn
weaker black
strong white
--
black first
 advantage
 white
   7.5 points extra
--
intersection place
even on edge
even on corner
---
most territory win
---
points
 capture

 supply line cut off  -> die
 prisonor goes bowl

 ---
 capture simple
--
corner capture
 2 supply line

 side
  3 supply line

center
 4supply line

---
no diagonal
---
some shape strong, some shape weak
---
2 rules only

 ko rule
   else where
 suicide role
   no suicide

  ---
  cuts of every body liberty
 ---
 saki
 ---
 trouble in corner
 saki pattern

--

open ai
learning dexterity

roate block in any orientation
human hand
---
domain randomization

fast, heavy, color, friction
---
rapid
rollout workers
optimizers
---
rotate other shape

sound , light
---
open ai five

christy dennison
machine learning engineer, open ai
---
made the bot
this year
new ai system
last
reinforcement learnign
playing against amateur
---
william lee(blitz)
kil lthe enemy building, take tam
spells
---
strategy
self play
100,000 cpu
self learn
---
hyper-parameter team spirit
team maets
--
playas one unit
---
best player in audience
---
nudge block
graps
puck
---
base line
---
learn from failure
gym
her
---
safe artificail intelligence
---
self play
---
international dota champions
---
bot astrainging routine
---
learned something
this will happen
---
some tell
experience a





0---
machine learning
deep learning
data scient
artificial intelligence
---
appliedaicourse
---
google deep mind
demiss hassbis
---
ancient board game
---
learn on their own
--
learn & truely learn
---
why is this important
machine can easy learn
---
previous ai
 brutforce
---
reinforment
---
go have more combination
---
learn without alteration
---
generic general purpose learn
---
it felt right
---
intuitive right
not logic based
---
google deep mind
2010 british mind
 500 million dollar
---
invest to keep eye on them
---

demis
 ches prodigy
 lion head
  black & white
 london
  future co founder of deep mind

goal
 solve intelligence
 machine
 power general purose learning
 formalize intelinge
 how it
 understand human brain

---
deep inforcement learning
1lakh
 internet
 mimic human player
 stronger
 compete 
 play 30 million times
 ---

 incrementally avoid errors
 ---
 neural network
 acces external memory
 ---
 micro intelligence
 --
 learn from own experince
 space invaders
 ---
 controlling
 after training
 overnight
 normal computer 8 hours
 ---
 predictive shots
 after that
 ---
 more games
 ---
 1970, 80
 complex 3d games
 doom
 ---
 health care, science
 speed in break through
 helping human expoert improve more
 ---

 computer vision
 general computer
 news publishing
 news writing
 ===
 digital reinforce
 ---
 few of unknown
 ---
 ant, bird, chimp, dumb human, eintesin
 ---
 reservation stephen, elon musk
 paul collen
 ---
 cautiously optimist
 deepmand hq london
 ---
 ai ethics board
 ---
 memory & imaginations
 2007
 ---
 https://www.linkedin.com/search/results/people/?firstName=&lastName=Hassabis&origin=SEO_SN



---
machine learning amity one
---