What are Convolutional Neural Networks?(10 points)
For understanding the writings of this post, you should have idea of machine learning, deep learning and its basic terms, you can get mine one here.
- A Convolutional Neural Network(CNN or ConvNet) is a class of deep neural networks, most commonly applied to analyzing visual imagery.
2. CNNs are inspired by the biological processes in that connectivity pattern between neurons resembles the organization of the animal visual cortex.
3. Also known as shift invariant or space invariant artificial neural networks(SIANN), based on their shared-weights architecture and translation in variance characteristics.
4. CNNs are regularized versions of multilayer perceptrons. MLPs are fully connected networks, which makes them prone to overfitting.
5. The difference is that CNNs take a different approach of regularization- they take the advantage of the hierarchical pattern data and assemble more complex patterns using smaller and simpler patterns. Therefore , the scale of connectedness and complexity, CNNs are on the lower extreme.
6. CNNs use pre-processing compared to other image classification algorithms, that is, the network learns the filters that in traditional algorithms were hand-engineered.
7. CNN indicates that the network employs a mathematical operation called convolution which is a specialized kind of linear operation.
8. CNNs use convolution in place of general matrix multiplication at least in one layer.
9. The basic CNN architecture consists of: Input->(Conv+ReLU)->Pooling ->(Conv+ReLU)->Pooling-> Flatten->Fully Connected-> Softmax-> Output
10. The feature extraction is carried out in the Convolutional layer+ReLU and Pooling layers and the classification is carried out in Fully Connected and Softmax layers.
Know more about architecture of CNN here.
For queries, feel free to write in the comment💬 section below. You can connect with me on LinkedIn !!
Thank you for reading! Have a great day ahead😃
