SlideShare a Scribd company logo
1 of 17
How to denote general computation in neural networks
1
 :the width of
 :the th layer.  :an activation
function
A simple example of layers of a neural net
2
 A simple
example of
layers of a
neural net
Chain rules and computation graphs
3
 A chain rule without junctions
 A chain rule with a junction
•
•
•
•
Chain rules and computation graphs
4
 General chain rules
•
•
Partial differentiations of a loss function with respect to parameters

5
A chain rule
without junctions
Partial differentiations of a loss function with respect to parameters
 The same computation with respect to
all the parameters
6
•
:Error
•
Calculating errors

7
Chain rule
with a junction
Computation of errors
 Calculating errors with vectors and matrices
8
Recursive computation
Generalizing the computation
9
Can be
generalized
layers with
any size
 
Recursiveness of backprop
10
Error function
Backprop


Forward backprop


boldsymbol{a}_t = W_{t-1}boldsymbol{o}_{t-1}
boldsymbol{o}_t = boldsymbol{sigma}( boldsymbol{a}_{t}) left(
begin{array}{c}
a_{t, 1} 
a_{t, 2}
end{array}
right) = left(
begin{array}{ccc}
w_{1, 1}&w_{1, 2}&w
w_{2, 1}&w_{2, 2} &
end{array}
right)left(
begin{array}{c}
o_{t-1,1} 
o_{t-1,2} 
o_{t-1,3}
end{array}
right)
left(
begin{array}{c}
a_{t+1, 1}
a_{t+1, 2}
frac{partial ell}{partialw_{1, 1}} = frac{partial ell}{partialo_{t, 1}}
frac{partial o_{t, 1}}{partialw_{1, 1}}
= frac{partial ell}{partial o_{t, 1}}frac{partial sigma (a_{t, 1})}{partialw_{1, 1}}
= frac{partial ell}{partial o_{t, 1}}frac{partial sigma (a_{t, 1})}{partiala_{t, 1}}
frac{partial a_{t, 1}}{partialw_{1, 1}}
= frac{partial ell}{partial o_{t, 1}} sigma ' (a_{t, 1}) o_{t-1,1}
frac{partial ell}{partialw_{1, 1}} = frac{partial ell}{partialo_{t, 1}} sigm
(a_{t, 1})o_{t-1,1}
left(
begin{array}{ccc}
frac{partial ell}{partialw_{1, 1}} &frac{partial ell}{partial w_{1,2}} &
frac{partial ell}{partialw_{1, 3}} 
frac{partial ell}{partialw_{2, 1}} &frac{partial ell}{partial w_{2,
2}} & frac{partial ell}{partialw_{2,3}} 
end{array}
right)=left(
begin{array}{ccc}
frac{partial ell}{partialo_{t, 1}} sigma ' (a_{t, 1})o_{t-1,1} &frac{partia
ell}{partial o_{t, 1}} sigma ' (a_{t, 1}) o_{t-1,2} &frac{partial ell}{partial
1}} sigma '(a_{t, 1})o_{t-1,3} 
frac{partial ell}{partialo_{t, 2}} sigma ' (a_{t, 2})o_{t-1,1} &frac{partia
ell}{partial o_{t, 2}} sigma ' (a_{t, 2}) o_{t-1,2} &frac{partial ell}{partial
2}} sigma '(a_{t, 2}) o_{t-1,3} 
end{array}
right)
right)= left(
begin{array}{cc}
sigma '(a_{t, 1}) &0
0 & sigma '(a_{t, 2}) 
end{array}
right)left(
begin{array}{c}
frac{partial ell}{partialo_{t, 1}} 
frac{partial ell}{partialo_{t, 2}}
end{array}
right) left(
begin{array}{ccc}
o_{t-1,1} & o_{t-1,2} & o_{t-1,3} 
end{array}
right)
= diag(boldsymbol{sigma}' (boldsymbol{a}_t) )
boldsymbol{delta_t}boldsymbol{o}_{t-1}^T
boldsymbol{delta_t} = frac{partial ell}{partial boldsymbol{o}_t}
= frac{partial ell}{partial o_{t+1, 1}} frac{partialo_{t+1, 1}}{parti
1}} frac{partial a_{t+1, 1}}{partialo_{t, 1}}+ cdots +frac{partial e
o_{t+1, 4}} frac{partial o_{t+1, 4}}{partiala_{t+1,4}} frac{partiala
4}}{partialo_{t, 1}}
= frac{partial ell}{partial o_{t+1, 1}} fra
a_{t+1, 1}} frac{partiala_{t+1, 1}}{partia
ell}{partial o_{t+1,4}} frac{partial sigm
4}} frac{partial a_{t+1, 4}}{partialo_{t, 1
= frac{partial ell}{partial o_{t+1, 1}} sigma ' (a_{t+1, 1}) phi_{1,
de
de
en
rig
be
ph
ph
ph
ph
en
rig
sig
le
be
de
de
de
de
en
rig
Brief instruction on backprop

More Related Content

What's hot

Functions and formulas
Functions and formulasFunctions and formulas
Functions and formulasFrya Lora
 
Stack & Queue using Linked List in Data Structure
Stack & Queue using Linked List in Data StructureStack & Queue using Linked List in Data Structure
Stack & Queue using Linked List in Data StructureMeghaj Mallick
 
Exercism Challenge - Saddle Point
Exercism Challenge - Saddle PointExercism Challenge - Saddle Point
Exercism Challenge - Saddle PointPatrick Dunn
 
Array sorting
Array sortingArray sorting
Array sortingALI RAZA
 
Adbms 39 algorithms for project and set operations
Adbms 39 algorithms for project and set operationsAdbms 39 algorithms for project and set operations
Adbms 39 algorithms for project and set operationsVaibhav Khanna
 
Matlab operators
Matlab operatorsMatlab operators
Matlab operatorsAswin Pv
 
Counting sort(Non Comparison Sort)
Counting sort(Non Comparison Sort)Counting sort(Non Comparison Sort)
Counting sort(Non Comparison Sort)Hossain Md Shakhawat
 
Algorithms Lecture 8: Pattern Algorithms
Algorithms Lecture 8: Pattern AlgorithmsAlgorithms Lecture 8: Pattern Algorithms
Algorithms Lecture 8: Pattern AlgorithmsMohamed Loey
 
Code tuning techniques
Code tuning techniquesCode tuning techniques
Code tuning techniquesAsha Sari
 
c++ programming Unit 3 variables,data types
c++ programming Unit 3 variables,data typesc++ programming Unit 3 variables,data types
c++ programming Unit 3 variables,data typesAAKASH KUMAR
 
Lecture 10 11-signal_flow_graphs
Lecture 10 11-signal_flow_graphsLecture 10 11-signal_flow_graphs
Lecture 10 11-signal_flow_graphsSaifullah Memon
 

What's hot (20)

Insertion and merge sort
Insertion and merge sortInsertion and merge sort
Insertion and merge sort
 
Functions and formulas
Functions and formulasFunctions and formulas
Functions and formulas
 
Lecture 7
Lecture 7Lecture 7
Lecture 7
 
Stack & Queue using Linked List in Data Structure
Stack & Queue using Linked List in Data StructureStack & Queue using Linked List in Data Structure
Stack & Queue using Linked List in Data Structure
 
Exercism Challenge - Saddle Point
Exercism Challenge - Saddle PointExercism Challenge - Saddle Point
Exercism Challenge - Saddle Point
 
Array sorting
Array sortingArray sorting
Array sorting
 
Adbms 39 algorithms for project and set operations
Adbms 39 algorithms for project and set operationsAdbms 39 algorithms for project and set operations
Adbms 39 algorithms for project and set operations
 
Merge sort
Merge sortMerge sort
Merge sort
 
Matlab matrics
Matlab matricsMatlab matrics
Matlab matrics
 
Matlab operators
Matlab operatorsMatlab operators
Matlab operators
 
Counting sort(Non Comparison Sort)
Counting sort(Non Comparison Sort)Counting sort(Non Comparison Sort)
Counting sort(Non Comparison Sort)
 
3 Array operations
3   Array operations3   Array operations
3 Array operations
 
Graph theory
Graph theoryGraph theory
Graph theory
 
R time series analysis
R   time series analysisR   time series analysis
R time series analysis
 
Algorithms Lecture 8: Pattern Algorithms
Algorithms Lecture 8: Pattern AlgorithmsAlgorithms Lecture 8: Pattern Algorithms
Algorithms Lecture 8: Pattern Algorithms
 
Code tuning techniques
Code tuning techniquesCode tuning techniques
Code tuning techniques
 
Adv excel
Adv excelAdv excel
Adv excel
 
c++ programming Unit 3 variables,data types
c++ programming Unit 3 variables,data typesc++ programming Unit 3 variables,data types
c++ programming Unit 3 variables,data types
 
Lecture 10 11-signal_flow_graphs
Lecture 10 11-signal_flow_graphsLecture 10 11-signal_flow_graphs
Lecture 10 11-signal_flow_graphs
 
Merge sort-algorithm for computer science engineering students
Merge sort-algorithm for computer science engineering studentsMerge sort-algorithm for computer science engineering students
Merge sort-algorithm for computer science engineering students
 

Similar to Brief instruction on backprop

Chem modiagram en
Chem modiagram enChem modiagram en
Chem modiagram enHerberSoria
 
"SSumM: Sparse Summarization of Massive Graphs", KDD 2020
"SSumM: Sparse Summarization of Massive Graphs", KDD 2020"SSumM: Sparse Summarization of Massive Graphs", KDD 2020
"SSumM: Sparse Summarization of Massive Graphs", KDD 2020KyuhanLee4
 
symbols of Latex.pdf
symbols of Latex.pdfsymbols of Latex.pdf
symbols of Latex.pdfMdNaim23
 
Latex Symbols.pdf
Latex Symbols.pdfLatex Symbols.pdf
Latex Symbols.pdfKomal526846
 
EE380-4 Course project Experimental determination of a Ser.docx
EE380-4  Course project Experimental determination of a  Ser.docxEE380-4  Course project Experimental determination of a  Ser.docx
EE380-4 Course project Experimental determination of a Ser.docxjack60216
 
2014-mo444-practical-assignment-04-paulo_faria
2014-mo444-practical-assignment-04-paulo_faria2014-mo444-practical-assignment-04-paulo_faria
2014-mo444-practical-assignment-04-paulo_fariaPaulo Faria
 
GraphBLAS: A linear algebraic approach for high-performance graph queries
GraphBLAS: A linear algebraic approach for high-performance graph queriesGraphBLAS: A linear algebraic approach for high-performance graph queries
GraphBLAS: A linear algebraic approach for high-performance graph queriesGábor Szárnyas
 
Electrical Engineering Assignment Help
Electrical Engineering Assignment HelpElectrical Engineering Assignment Help
Electrical Engineering Assignment HelpEdu Assignment Help
 
Unit-1 Basic Concept of Algorithm.pptx
Unit-1 Basic Concept of Algorithm.pptxUnit-1 Basic Concept of Algorithm.pptx
Unit-1 Basic Concept of Algorithm.pptxssuser01e301
 
Introduction to matlab
Introduction to matlabIntroduction to matlab
Introduction to matlabVidhyaSenthil
 
Introduction to MatLab programming
Introduction to MatLab programmingIntroduction to MatLab programming
Introduction to MatLab programmingDamian T. Gordon
 
Direct solution of sparse network equations by optimally ordered triangular f...
Direct solution of sparse network equations by optimally ordered triangular f...Direct solution of sparse network equations by optimally ordered triangular f...
Direct solution of sparse network equations by optimally ordered triangular f...Dimas Ruliandi
 
QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...
QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...
QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...Austin Benson
 

Similar to Brief instruction on backprop (20)

Chem modiagram en
Chem modiagram enChem modiagram en
Chem modiagram en
 
La tex symbols
La tex symbolsLa tex symbols
La tex symbols
 
La tex symbols
La tex symbolsLa tex symbols
La tex symbols
 
Symbols
SymbolsSymbols
Symbols
 
ECG BASED REPORT.
ECG BASED REPORT.ECG BASED REPORT.
ECG BASED REPORT.
 
"SSumM: Sparse Summarization of Massive Graphs", KDD 2020
"SSumM: Sparse Summarization of Massive Graphs", KDD 2020"SSumM: Sparse Summarization of Massive Graphs", KDD 2020
"SSumM: Sparse Summarization of Massive Graphs", KDD 2020
 
symbols of Latex.pdf
symbols of Latex.pdfsymbols of Latex.pdf
symbols of Latex.pdf
 
Latex Symbols.pdf
Latex Symbols.pdfLatex Symbols.pdf
Latex Symbols.pdf
 
EE380-4 Course project Experimental determination of a Ser.docx
EE380-4  Course project Experimental determination of a  Ser.docxEE380-4  Course project Experimental determination of a  Ser.docx
EE380-4 Course project Experimental determination of a Ser.docx
 
2014-mo444-practical-assignment-04-paulo_faria
2014-mo444-practical-assignment-04-paulo_faria2014-mo444-practical-assignment-04-paulo_faria
2014-mo444-practical-assignment-04-paulo_faria
 
GraphBLAS: A linear algebraic approach for high-performance graph queries
GraphBLAS: A linear algebraic approach for high-performance graph queriesGraphBLAS: A linear algebraic approach for high-performance graph queries
GraphBLAS: A linear algebraic approach for high-performance graph queries
 
Curvefitting
CurvefittingCurvefitting
Curvefitting
 
Basic concepts in_matlab
Basic concepts in_matlabBasic concepts in_matlab
Basic concepts in_matlab
 
Electrical Engineering Assignment Help
Electrical Engineering Assignment HelpElectrical Engineering Assignment Help
Electrical Engineering Assignment Help
 
ECG
ECGECG
ECG
 
Unit-1 Basic Concept of Algorithm.pptx
Unit-1 Basic Concept of Algorithm.pptxUnit-1 Basic Concept of Algorithm.pptx
Unit-1 Basic Concept of Algorithm.pptx
 
Introduction to matlab
Introduction to matlabIntroduction to matlab
Introduction to matlab
 
Introduction to MatLab programming
Introduction to MatLab programmingIntroduction to MatLab programming
Introduction to MatLab programming
 
Direct solution of sparse network equations by optimally ordered triangular f...
Direct solution of sparse network equations by optimally ordered triangular f...Direct solution of sparse network equations by optimally ordered triangular f...
Direct solution of sparse network equations by optimally ordered triangular f...
 
QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...
QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...
QR Factorizations and SVDs for Tall-and-skinny Matrices in MapReduce Architec...
 

More from YasutoTamura1

How to formulate reinforcement learning in illustrative ways
How to formulate reinforcement learning in illustrative waysHow to formulate reinforcement learning in illustrative ways
How to formulate reinforcement learning in illustrative waysYasutoTamura1
 
Reinforcement course material samples: lecture 1
Reinforcement course material samples: lecture 1Reinforcement course material samples: lecture 1
Reinforcement course material samples: lecture 1YasutoTamura1
 
NLP_deep_learning_intro.pptx
NLP_deep_learning_intro.pptxNLP_deep_learning_intro.pptx
NLP_deep_learning_intro.pptxYasutoTamura1
 
Illustrative Introductory CNN
Illustrative Introductory CNNIllustrative Introductory CNN
Illustrative Introductory CNNYasutoTamura1
 
Illustrative Introductory Neural Networks
Illustrative Introductory Neural NetworksIllustrative Introductory Neural Networks
Illustrative Introductory Neural NetworksYasutoTamura1
 
Precise LSTM Algorithm
Precise LSTM AlgorithmPrecise LSTM Algorithm
Precise LSTM AlgorithmYasutoTamura1
 
simple_rnn_forward_back_propagation
simple_rnn_forward_back_propagationsimple_rnn_forward_back_propagation
simple_rnn_forward_back_propagationYasutoTamura1
 

More from YasutoTamura1 (8)

How to formulate reinforcement learning in illustrative ways
How to formulate reinforcement learning in illustrative waysHow to formulate reinforcement learning in illustrative ways
How to formulate reinforcement learning in illustrative ways
 
Reinforcement course material samples: lecture 1
Reinforcement course material samples: lecture 1Reinforcement course material samples: lecture 1
Reinforcement course material samples: lecture 1
 
NLP_deep_learning_intro.pptx
NLP_deep_learning_intro.pptxNLP_deep_learning_intro.pptx
NLP_deep_learning_intro.pptx
 
RL_in_10_min.pptx
RL_in_10_min.pptxRL_in_10_min.pptx
RL_in_10_min.pptx
 
Illustrative Introductory CNN
Illustrative Introductory CNNIllustrative Introductory CNN
Illustrative Introductory CNN
 
Illustrative Introductory Neural Networks
Illustrative Introductory Neural NetworksIllustrative Introductory Neural Networks
Illustrative Introductory Neural Networks
 
Precise LSTM Algorithm
Precise LSTM AlgorithmPrecise LSTM Algorithm
Precise LSTM Algorithm
 
simple_rnn_forward_back_propagation
simple_rnn_forward_back_propagationsimple_rnn_forward_back_propagation
simple_rnn_forward_back_propagation
 

Recently uploaded

Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticscarlostorres15106
 
Benefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other FrameworksBenefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other FrameworksSoftradix Technologies
 
"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr Bagan"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr BaganFwdays
 
Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!Commit University
 
My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024The Digital Insurer
 
Human Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsHuman Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsMark Billinghurst
 
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Enterprise Knowledge
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsMemoori
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebUiPathCommunity
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraDeakin University
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Mattias Andersson
 
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii SoldatenkoFwdays
 
Install Stable Diffusion in windows machine
Install Stable Diffusion in windows machineInstall Stable Diffusion in windows machine
Install Stable Diffusion in windows machinePadma Pradeep
 
Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfAlex Barbosa Coqueiro
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationRidwan Fadjar
 
Streamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupStreamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupFlorian Wilhelm
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...shyamraj55
 

Recently uploaded (20)

Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
 
Benefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other FrameworksBenefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other Frameworks
 
"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr Bagan"ML in Production",Oleksandr Bagan
"ML in Production",Oleksandr Bagan
 
Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!
 
My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024
 
Human Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR SystemsHuman Factors of XR: Using Human Factors to Design XR Systems
Human Factors of XR: Using Human Factors to Design XR Systems
 
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024
 
Vulnerability_Management_GRC_by Sohang Sengupta.pptx
Vulnerability_Management_GRC_by Sohang Sengupta.pptxVulnerability_Management_GRC_by Sohang Sengupta.pptx
Vulnerability_Management_GRC_by Sohang Sengupta.pptx
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial Buildings
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio Web
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning era
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?
 
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko
 
Install Stable Diffusion in windows machine
Install Stable Diffusion in windows machineInstall Stable Diffusion in windows machine
Install Stable Diffusion in windows machine
 
Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdf
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 Presentation
 
Streamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project SetupStreamlining Python Development: A Guide to a Modern Project Setup
Streamlining Python Development: A Guide to a Modern Project Setup
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
 

Brief instruction on backprop

  • 1. How to denote general computation in neural networks 1  :the width of  :the th layer.  :an activation function
  • 2. A simple example of layers of a neural net 2  A simple example of layers of a neural net
  • 3. Chain rules and computation graphs 3  A chain rule without junctions  A chain rule with a junction • • • •
  • 4. Chain rules and computation graphs 4  General chain rules • •
  • 5. Partial differentiations of a loss function with respect to parameters  5 A chain rule without junctions
  • 6. Partial differentiations of a loss function with respect to parameters  The same computation with respect to all the parameters 6 • :Error •
  • 8. Computation of errors  Calculating errors with vectors and matrices 8 Recursive computation
  • 9. Generalizing the computation 9 Can be generalized layers with any size  
  • 10. Recursiveness of backprop 10 Error function Backprop   Forward backprop  
  • 11. boldsymbol{a}_t = W_{t-1}boldsymbol{o}_{t-1} boldsymbol{o}_t = boldsymbol{sigma}( boldsymbol{a}_{t}) left( begin{array}{c} a_{t, 1} a_{t, 2} end{array} right) = left( begin{array}{ccc} w_{1, 1}&w_{1, 2}&w w_{2, 1}&w_{2, 2} & end{array} right)left( begin{array}{c} o_{t-1,1} o_{t-1,2} o_{t-1,3} end{array} right) left( begin{array}{c} a_{t+1, 1} a_{t+1, 2}
  • 12. frac{partial ell}{partialw_{1, 1}} = frac{partial ell}{partialo_{t, 1}} frac{partial o_{t, 1}}{partialw_{1, 1}} = frac{partial ell}{partial o_{t, 1}}frac{partial sigma (a_{t, 1})}{partialw_{1, 1}} = frac{partial ell}{partial o_{t, 1}}frac{partial sigma (a_{t, 1})}{partiala_{t, 1}} frac{partial a_{t, 1}}{partialw_{1, 1}} = frac{partial ell}{partial o_{t, 1}} sigma ' (a_{t, 1}) o_{t-1,1}
  • 13. frac{partial ell}{partialw_{1, 1}} = frac{partial ell}{partialo_{t, 1}} sigm (a_{t, 1})o_{t-1,1} left( begin{array}{ccc} frac{partial ell}{partialw_{1, 1}} &frac{partial ell}{partial w_{1,2}} & frac{partial ell}{partialw_{1, 3}} frac{partial ell}{partialw_{2, 1}} &frac{partial ell}{partial w_{2, 2}} & frac{partial ell}{partialw_{2,3}} end{array} right)=left( begin{array}{ccc} frac{partial ell}{partialo_{t, 1}} sigma ' (a_{t, 1})o_{t-1,1} &frac{partia ell}{partial o_{t, 1}} sigma ' (a_{t, 1}) o_{t-1,2} &frac{partial ell}{partial 1}} sigma '(a_{t, 1})o_{t-1,3} frac{partial ell}{partialo_{t, 2}} sigma ' (a_{t, 2})o_{t-1,1} &frac{partia ell}{partial o_{t, 2}} sigma ' (a_{t, 2}) o_{t-1,2} &frac{partial ell}{partial 2}} sigma '(a_{t, 2}) o_{t-1,3} end{array} right)
  • 14. right)= left( begin{array}{cc} sigma '(a_{t, 1}) &0 0 & sigma '(a_{t, 2}) end{array} right)left( begin{array}{c} frac{partial ell}{partialo_{t, 1}} frac{partial ell}{partialo_{t, 2}} end{array} right) left( begin{array}{ccc} o_{t-1,1} & o_{t-1,2} & o_{t-1,3} end{array} right) = diag(boldsymbol{sigma}' (boldsymbol{a}_t) ) boldsymbol{delta_t}boldsymbol{o}_{t-1}^T boldsymbol{delta_t} = frac{partial ell}{partial boldsymbol{o}_t}
  • 15. = frac{partial ell}{partial o_{t+1, 1}} frac{partialo_{t+1, 1}}{parti 1}} frac{partial a_{t+1, 1}}{partialo_{t, 1}}+ cdots +frac{partial e o_{t+1, 4}} frac{partial o_{t+1, 4}}{partiala_{t+1,4}} frac{partiala 4}}{partialo_{t, 1}} = frac{partial ell}{partial o_{t+1, 1}} fra a_{t+1, 1}} frac{partiala_{t+1, 1}}{partia ell}{partial o_{t+1,4}} frac{partial sigm 4}} frac{partial a_{t+1, 4}}{partialo_{t, 1 = frac{partial ell}{partial o_{t+1, 1}} sigma ' (a_{t+1, 1}) phi_{1,