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
Theano: обучаем нейросеть, Екатерина Тузова, Py...
Search
IT-People
July 25, 2016
Programming
1
210
Theano: обучаем нейросеть, Екатерина Тузова, PyCharm
Выступление на конференции PyCon Russia 2016
IT-People
July 25, 2016
Tweet
Share
More Decks by IT-People
See All by IT-People
Александр Кошкин, Positive Technologies: Знай и люби свой CPython во имя луны и великой справедливости
itpeople123
1
200
How Python was Shaped by leaky Internals, Armin Ronacher, Flask framework
itpeople123
1
230
Building social network with Neo4j and Python, Андрей Солдатенко, Toptal
itpeople123
1
230
Осторожно, DSL! Иван Цыганов, Positive Technologies
itpeople123
0
120
Докеризация веб-приложения на Python Антон Егоров, Ostrovok.ru
itpeople123
0
190
Python Robot Operating System вместо Middleware framework Дмитрий Киселев, Endpoint.com
itpeople123
0
100
Falcon — очередная WSGI библиотека? Дмитрий Кукушкин, Wargaming
itpeople123
0
270
Очереди задач без купюр, Артем Малышев, Positive Technologies
itpeople123
1
110
Flame graph: новый взгляд на привычное профилирование, Кирилл Борисов, Яндекс
itpeople123
1
260
Other Decks in Programming
See All in Programming
Go言語はstack overflowの夢を見るか?
logica0419
0
520
The Past, Present, and Future of Enterprise Java
ivargrimstad
0
560
CSC305 Lecture 09
javiergs
PRO
0
300
TransformerからMCPまで(現代AIを理解するための羅針盤)
mickey_kubo
5
2.8k
Reactive Thinking with Signals and the Resource API
manfredsteyer
PRO
0
100
その面倒な作業、「Dart」にやらせませんか? Flutter開発者のための業務効率化
yordgenome03
1
140
Android16 Migration Stories ~Building a Pattern for Android OS upgrades~
reoandroider
0
130
20251016_Rails News ~Rails 8.1の足音を聴く~
morimorihoge
2
670
Devvox Belgium - Agentic AI Patterns
kdubois
1
140
Writing Better Go: Lessons from 10 Code Reviews
konradreiche
3
5.7k
Domain-centric? Why Hexagonal, Onion, and Clean Architecture Are Answers to the Wrong Question
olivergierke
3
940
あなたとKaigi on Rails / Kaigi on Rails + You
shimoju
0
180
Featured
See All Featured
The Power of CSS Pseudo Elements
geoffreycrofte
79
6k
Code Reviewing Like a Champion
maltzj
526
40k
Stop Working from a Prison Cell
hatefulcrawdad
271
21k
CoffeeScript is Beautiful & I Never Want to Write Plain JavaScript Again
sstephenson
162
15k
Music & Morning Musume
bryan
46
6.9k
Keith and Marios Guide to Fast Websites
keithpitt
411
23k
The Cult of Friendly URLs
andyhume
79
6.6k
How to Ace a Technical Interview
jacobian
280
24k
Embracing the Ebb and Flow
colly
88
4.9k
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
PRO
23
1.5k
実際に使うSQLの書き方 徹底解説 / pgcon21j-tutorial
soudai
PRO
190
55k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
35
3.2k
Transcript
Theano: let’s build a net Ekaterina Tuzova
Motivation
MNIST dataset X — images (60000, 784) Y — labels
(60000, ) Goal: train a model to look at images and predict what digits they are.
K nearest neighbors
K nearest neighbors ~95 % accuracy
Feature engineering
Machine Learning
Machine Learning
Variable in Computer Science A piece of informations stored somewhere
in memory with a name made up by a programmer.
Variable in Mathematics Indeterminate quantity that can be manipulated using
arithmetic operations, or using calculus and so on.
Theano
What is Theano? A library for efficiently handling mathematical expressions
involving tensors.
What is tensor? Multi-dimensional array.
What is Theano? A mathematical symbolic expression compiler.
Quick How-To - Define symbolic expression - Compile a function
that can compute numeric values using those expressions - Execute that function on data
Example model GoogLeNet, 2014 ~99.6% accuracy
Logistic regression Probabilistic, linear classifier.
Simple model
Preparation
Load MNIST
Define model
Theano shared variable Shared between different functions and function calls.
Build model
Train model
Train model ~93% accuracy
Simple model
Improve model
Improve model
Improve model
Train model ~98.5% accuracy
Thank you. @ktisha