After the first CNN-based architecture (AlexNet) that win the ImageNet 2012 competition, Every subsequent winning architecture uses more layers in a deep neural network to reduce the error rate. This works for less number of layers, but when we increase the number of layers, there is a common problem in deep learning associated with that called Vanishing/Exploding gradient. This causes the gradient to become 0 or too large. Thus when we increases number of layers, the training and test error rate also increases.
In the above plot, we can observe that a 56-layer CNN gives more error rate on both training and testing dataset than a 20-layer CNN architecture, If this was the result of over fitting, then we should have lower training error in 56-layer CNN but then it also has higher training error. After analyzing more on error rate the authors were able to reach conclusion that it is caused by vanishing/exploding gradient.
ResNet, which was proposed in 2015 by researchers at Microsoft Research introduced a new architecture called Residual Network.
In order to solve the problem of the vanishing/exploding gradient, this architecture introduced the concept called Residual Network. In this network we use a technique called skip connections . The skip connection skips training from a few layers and connects directly to the output.
The approach behind this network is instead of layers learn the underlying mapping, we allow network fit the residual mapping. So, instead of say H(x), initial mapping, let the network fit, F(x) := H(x) – x which gives H(x) := F(x) + x.
The advantage of adding this type of skip connection is because if any layer hurt the performance of architecture then it will be skipped by regularization. So, this results in training very deep neural network without the problems caused by vanishing/exploding gradient. The authors of the paper experimented on 100-1000 layers on CIFAR-10 dataset.
There is a similar approach called “highway networks”, these networks also uses skip connection. Similar to LSTM these skip connections also uses parametric gates. These gates determine how much information passes through the skip connection. This architecture however has not provide accuracy better than ResNet architecture.
This network uses a 34-layer plain network architecture inspired by VGG-19 in which then the shortcut connection is added. These shortcut connections then convert the architecture into residual network.
Using the Tensorflow and Keras API, we can design ResNet architecture (including Residual Blocks) from scratch. Below is the implementation of different ResNet architecture. For this implementation we use CIFAR-10 dataset. This dataset contains 60, 000 32×32 color images in 10 different classes (airplanes, cars, birds, cats, deer, dogs, frogs, horses, ships, and trucks) etc. This datasets can be assessed from keras.datasets API function.
- First, we import the keras module and its APIs. These APIs help in building architecture of the ResNet model.
- Now, We set different hyper parameters that is required for ResNet architecture. We also done some preprocess our datasets to prepare it for training.
- In this step, we set the learning rate according to the number of epochs. As the number of epochs the learning rate must be decreased to ensure better learning.
Code: Setting LR for different number of Epochs
- In this step we define basic ResNet building block that can be used for defining the ResNet V1 and V2 architecture.
- In this step we define ResNet V1 architecture that is based on the ResNet building block we defined above:
- In this step we define ResNet V2 architecture that is based on the ResNet building block we defined above:
- The code below is used to train and test the ResNet v1 and v2 architecture we defined above:
Results & Conclusion:
On the ImageNet dataset, the authors uses a 152-layers ResNet, which is 8 times more deep than VGG19 but still have less parameters. An ensemble of these ResNets generated an error of only 3.7% on ImageNet test set, the result which won ILSVRC 2015 competition. On COCO object detection dataset, it also generates a 28% relative improvement due to its very deep representation.
- The result above shows that shortcut connections would be able to solve the problem caused by increasing the layers because as we increase layers from 18 to 34 the error rate on ImageNet Validation Set also decreases unlike the plain network.
- Below are the results on ImageNet Test Set. The 3.57% top-5 error rate of ResNet was the lowest and thus ResNet architecture came first in ImageNet classification challenge in 2015.
- Introduction to Residual Networks
- Introduction to Multi-Task Learning(MTL) for Deep Learning
- Artificial intelligence vs Machine Learning vs Deep Learning
- Difference Between Artificial Intelligence vs Machine Learning vs Deep Learning
- Difference Between Machine Learning and Deep Learning
- Need of Data Structures and Algorithms for Deep Learning and Machine Learning
- Introduction to Deep Learning
- ML | Natural Language Processing using Deep Learning
- Deep Learning | Introduction to Long Short Term Memory
- Deep Learning with PyTorch | An Introduction
- Prediction of Wine type using Deep Learning
- Avengers Endgame and Deep learning | Image Caption Generation using the Avengers EndGames Characters
- Deep Q-Learning
- Implementing Deep Q-Learning using Tensorflow
- Differential Privacy and Deep Learning
- Human Activity Recognition - Using Deep Learning Model
- ML - List of Deep Learning Layers
- ML - Saving a Deep Learning model in Keras
- DLSS - Deep Learning Super Sampling
- Computational Graphs in Deep Learning
If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to email@example.com. See your article appearing on the GeeksforGeeks main page and help other Geeks.
Please Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.