EPSRC Winter School in Mathematics for Data Modelling

The Multi-layer Perceptron

author: Robert F Harrison, University of Sheffield

Description

This presentation describes the multilayer perceptron and practical issues in data modelling.

You might be experiencing some problems with Your Video player.

*Slides*
0:00	Non-Linear Modelling by Adaptive Pre-Processing
1:28	The Data Modelling Problem - 1
7:20	The Data Modelling Problem - 2
7:39	- Questions
7:45	The Data Modelling Problem - 2
8:10	The Data Modelling Problem - 3
8:46	The Data Modelling Problem - 4
9:18	The Data Modelling Problem - 5
9:32	The Data Modelling Problem - 6
10:19	The Data Modelling Problem - 7
10:40	Dimensionality
15:42	What Goes on in the Gaps?
18:05	Overfitting (Sample Data)
19:27	Underfitting (Sample Data)
19:55	Goldilocks
20:44	Restricting “Flexibility”
29:26	Hold-Out Method
34:44	Cross Validation - 1
38:29	Cross Validation - 2
39:48	Cross Validation - 3
39:55	Cross Validation - 4
40:31	Adaptive Basis Functions - 1
40:35	- Questions
41:40	Adaptive Basis Functions - 1
44:27	Adaptive Basis Functions - 2
45:39	The Multi-Layer Perceptron
48:27	Two-Layer MLP
49:53	A Sigmoidal Unit
50:43	Combinations of Sigmoids - 1
52:08	Combinations of Sigmoids - 2
52:44	Universal Approximation
55:42	Interpretation
62:08	Pros
64:10	Compactness of Model
66:46	Backpropagation Algorithm
67:08	& Cons
67:27	- Questions
69:24	& Cons
70:20	Rolling Ball
71:19	Gradient Descent - 1
72:15	Gradient Descent - 2
74:11	Multi-Modal Cost Service
74:27	Heading Downhill - 1
75:58	Heading Downhill - 2
76:19	Heading Downhill - 3
76:38	Heading Downhill - 4
77:18	Implications
80:14	RBF NN Warning!
81:39	Are Multiple Minima a Problem?
83:01	How to Use
85:10	Local Solutions

Lecture rating

People found this lecture:
Worth seeing
because it is:
Valuable and informative
Well presented
Easily understandable
Acceptably recorded
You need to login to cast your vote.

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.