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
Baby Machine Learning
Search
Buzzvil
June 23, 2021
Programming
0
130
Baby Machine Learning
By Hes
Buzzvil
June 23, 2021
Tweet
Share
More Decks by Buzzvil
See All by Buzzvil
220903_GFS
buzzvil
0
330
Git 해부하기 2 + 3
buzzvil
0
16
Metastable Failure
buzzvil
0
140
Git 해부하기
buzzvil
0
26
Introduction to Plate Solving
buzzvil
0
9
Airbnb Minerva
buzzvil
0
180
Shape up 방법론
buzzvil
0
790
Buzzvil Billing Data Pipeline
buzzvil
0
410
Journey of Dash's release-cycle
buzzvil
0
130
Other Decks in Programming
See All in Programming
サイコロで理解する統計的仮説検定の考え方
tatamiya
4
850
ゆるい個人開発のススメ
kuroppe1819
10
980
Rails と人魚の話/rails-and-mermaid
sanfrecce_osaka
0
100
Blue/Greenデプロイの導入による 運用フローの改善
kudoas
1
360
1BRC--Nerd Sniping the Java Community
gunnarmorling
0
340
Elm Form Validation
bkuhlmann
0
510
MetricKitで予期せぬ終了を検知する話 / Detect unexpected termination with MetricKit
nekowen
0
180
Code Reviews
bkuhlmann
4
890
Git Rebase
bkuhlmann
11
1.6k
冗長なエラーログを削減し、スタックトレースを手に入れる / Reducing Verbose Error Logs and Obtaining Stack Traces
upamune
0
170
What We Can Learn From OSS
inouehi
0
420
try! Swift Tokyo 2024 参加報告 / try! Swift Tokyo 2024 Report
hironytic
0
200
Featured
See All Featured
The MySQL Ecosystem @ GitHub 2015
samlambert
242
12k
Mobile First: as difficult as doing things right
swwweet
216
8.6k
Atom: Resistance is Futile
akmur
258
25k
Scaling GitHub
holman
457
140k
Designing Experiences People Love
moore
136
23k
The Pragmatic Product Professional
lauravandoore
24
5.8k
Optimizing for Happiness
mojombo
370
69k
Distributed Sagas: A Protocol for Coordinating Microservices
caitiem20
321
20k
Cheating the UX When There Is Nothing More to Optimize - PixelPioneers
stephaniewalter
273
13k
YesSQL, Process and Tooling at Scale
rocio
163
13k
Embracing the Ebb and Flow
colly
79
4.1k
BBQ
matthewcrist
80
8.8k
Transcript
Hes 2021-06-23 Baby Machine Learning From Regression To Neural Network
Machine Learning •AI •Deep Learning •Classi fi cation •Regression •Clustering
•Data Mining •Reinforcement Learning ...umm "MATH"
Machine Learning •AI •Deep Learning •Classi fi cation •Regression •Clustering
•Data Mining •Reinforcement Learning ...umm "MATH"
How people think about House price • How many people?
• How Big? • How many rooms & bathrooms? • Where the house is? + ठࣁӂ, ࣁӂ
How people think about House price - Relation • How
many people? • Size • # Rooms & Bathrooms • Where the house is? • How Big? • ठࣁӂ • ࣁӂ • Family Size • Safety
How people think about House price - Priority • How
many people? • Size • # Rooms & Bathrooms • Where the house is? • How Big? • ठࣁӂ • ࣁӂ • Family Size • Safety 1
How people think about House price - Priority • How
many people? • Size • # Rooms & Bathrooms • Where the house is? • How Big? • ठࣁӂ • ࣁӂ • Family Size • Safety 1 Depends on a situation
Demand makes Price How people think about
Demand makes Price How people think about
makes Price Various Situation How people think about
Data Various Situation How people think about
Learn Method from Model variable Brain Data How people think
about
Let's start with simple model
What makes price Overview Location point • ठࣁӂ • ࣁӂ
• Safety • Family Size • Size • Rooms &BR Volume point Family point Price
What makes price Overview Location point • ठࣁӂ • ࣁӂ
• Safety • Family Size • Size • Rooms &BR Volume point Family point Price
Linear Regression House price • ठࣁӂ • ࣁӂ • Safety
? ? ? • Family Size ? Location Point
Linear Regression House price • ठࣁӂ • ࣁӂ • Safety
? ? ? • Family Size ? Location Point Safetyо 95, ठࣁӂ 1(True), ࣁӂ 0(False), Family Sizeо 4ݺݶ, Location pointח 77 ա৬ঠ ೞפө... ۧѱ ߄Լݶ!
Linear Regression House price • ठࣁӂ • ࣁӂ • Safety
? ? ? • Family Size ? Location Point Ӓېࢲ ӒѦ ־о ೞחؘ?! ஏ Ѿҗ৬ पઁ чਸ ৈח ੌਸ.. Optimizerо ೞݶ غѷ!!
Linear Regression House price • ठࣁӂ • ࣁӂ • Safety
? ? ? • Family Size ? Location Point Ӓېࢲ ӒѦ ־о ೞחؘ?! ஏ Ѿҗ৬ पઁ чਸ ৈח ੌਸ.. Optimizerо ೞݶ غѷ!!
Linear Regression Happy case
Linear Regression Umm.. case
Activation Fuction move on from Linear Linear Regression = Non-Linear
Result + Activation(nonlinearities)
Activation Fuction move on from Linear Linear Regression = Non-Linear
Result + Activation(nonlinearities)
σ( ) Location point Logistic Regression House price =
σ( ) Location point Logistic Regression House price • ठࣁӂ
• ࣁӂ • Safety ? ? ? • Family Size ? Good / Not Good (Location) = աח ݅ೡ ഛܫਸ ঌҊ रয.. Ӓېࢲ ഛܫ?
Neural Network House price - Stack the regression a( )
Location point • ठࣁӂ • ࣁӂ • Safety • Family Size • Size • Rooms &BR a( ) Volume point a( ) Family point Price
Neural Network House price - Stack the regression a( )
Location point • ठࣁӂ • ࣁӂ • Safety • Family Size • Size • Rooms &BR a( ) Volume point a( ) Family point Prob σ( )
Neural Network House price - Stack the regression a( )
Location point • ठࣁӂ • ࣁӂ • Safety • Family Size • Size • Rooms &BR a( ) Volume point a( ) Family point Prob σ( ) Prob σ( ) Prob σ( ) ARGMAX
What we can do • Probability of Conversion • Classi
fi cation • Scoring • Ranking …
ೞ݅ ৈ য۵..