Deep Residual Learning for Image Recognition

I am reading this paper because it was recommended as part of Ilya Sutskever's approx. 30 papers that he recommended to John Carmack to learn what really matters for machine learning / AI today. This paper "presents a residual training framework to ease the training of networks that are substantially deeper than those used previously".

Reference Link to PDF of Paper

Date Created:
Last Edited:
0 14

0.1 Abstract

Deep Neural Networks are more difficult to train. This paper presents a residual learning framework to ease the training of networks that are substantially deeper than those used previously. The paper reformulates layers as leaning residual functions with reference to the layer inputs, instead of learning unreferenced functions. This paper provides comprehensive empirical evidence that these residual networks are easier to optimize and can gain accuracy from considerably increased depth. The depth of representations is of central importance for many visual recognition tasks.

Deep convolutional neural networks have led to a series of breakthroughs for image classification. Deep networks naturally integrate low/mid/high level features and classifiers in an end-to-end multilayer fashion and the ”levels” of features can be enriched by the number of stacked layers. Depth of the network is of central importance in difficult tasks. The depth of networks was constrained by the exploding / vanishing gradients problem. This problem has been largely addressed ny normalized initialization and intermediate normalization layers, which enable networks with tens or layers to start converging for stochastic gradient descent (SGD) with backpropagation. When deeper networks are able to start converging, a *degradation* problem has been exposed: with the network depth increasing, accuracy gets saturated and then degrades rapidly - not caused by overfitting. This paper addresses the degradation problem by introducing a deep residual learning framework. Instead of hoping each few stacked layers directly fit a desired underlying mapping, we explicitly let these layers fit a residual mapping. This paper hypothesize that it is easier to optimize the residual mapping than to optimize the original, unreferenced mapping. The formulation of F(x) + x can be realized by feedforward neural networks with ”shortcut connections”. Shortcut connections are those skipping one or more layers. Shortcut connections in this case perform identity mapping, and their outputs are added to the outputs of the stacked layers. Identity shortcut connection add neither extra parmamaters nor computational complexity. The entire network can still by trained end-to-end by SGD with backpropagation, and can be easily implemented with common libraries without modifying the solvers.

PIC

Comments

You have to be logged in to add a comment

User Comments

Insert Math Markup

ESC
About Inserting Math Content
Display Style:

Embed News Content

ESC
About Embedding News Content

Embed Youtube Video

ESC
Embedding Youtube Videos

Embed TikTok Video

ESC
Embedding TikTok Videos

Embed X Post

ESC
Embedding X Posts

Embed Instagram Post

ESC
Embedding Instagram Posts

Insert Details Element

ESC

Example Output:

Summary Title
You will be able to insert content here after confirming the title of the <details> element.

Insert Table

ESC
Customization
Align:
Preview:

Insert Horizontal Rule

#000000

Preview:


View Content At Different Sizes

ESC

Edit Style of Block Nodes

ESC

Edit the background color, default text color, margin, padding, and border of block nodes. Editable block nodes include paragraphs, headers, and lists.

#ffffff
#000000

Edit Selected Cells

Change the background color, vertical align, and borders of the cells in the current selection.

#ffffff
Vertical Align:
Border
#000000
Border Style:

Edit Table

ESC
Customization:
Align:

Upload Lexical State

ESC

Upload a .lexical file. If the file type matches the type of the current editor, then a preview will be shown below the file input.

Upload 3D Object

ESC

Upload Jupyter Notebook

ESC

Upload a Jupyter notebook and embed the resulting HTML in the text editor.

Insert Custom HTML

ESC

Edit Image Background Color

ESC
#ffffff

Insert Columns Layout

ESC
Column Type:

Select Code Language

ESC
Select Coding Language

Insert Chart

ESC

Use the search box below

Upload Previous Version of Article State

ESC