@@ 1번째 줄: / 1번째 줄: @@
-== 노트 ==
-* The architecture of a CNN is designed to take advantage of the 2D structure of an input image (or other 2D input such as a speech signal).<ref name="ref_7639">[http://deeplearning.stanford.edu/tutorial/supervised/ConvolutionalNeuralNetwork/ Unsupervised Feature Learning and Deep Learning Tutorial]</ref>
-* A CNN consists of a number of convolutional and subsampling layers optionally followed by fully connected layers.<ref name="ref_7639" />
-* The figure below illustrates a full layer in a CNN consisting of convolutional and subsampling sublayers.<ref name="ref_7639" />
-* As the number of filters (output feature map depth) applied to the input increases, so does the number of features the CNN can extract.<ref name="ref_0526">[https://developers.google.com/machine-learning/practica/image-classification/convolutional-neural-networks ML Practicum: Image Classification]</ref>
-* Other CNNs may contain larger or smaller numbers of convolutional modules, and greater or fewer fully connected layers.<ref name="ref_0526" />
-* Besides, this article presents the fundamentals, tools, and two examples of the use of CNNs for fruit sorting and quality control.<ref name="ref_66bd">[https://www.mdpi.com/2076-3417/10/10/3443 A Review of Convolutional Neural Network Applied to Fruit Image Processing]</ref>
-* A convolutional neural network, or CNN, is a deep learning neural network designed for processing structured arrays of data such as images.<ref name="ref_d1ae">[https://deepai.org/machine-learning-glossary-and-terms/convolutional-neural-network Convolutional Neural Network]</ref>
-* It is this property that makes convolutional neural networks so powerful for computer vision.<ref name="ref_d1ae" />
-* The activation function has the effect of adding non-linearity into the convolutional neural network.<ref name="ref_d1ae" />
-* By the tenth layer, a convolutional neural network is able to detect more complex shapes such as eyes.<ref name="ref_d1ae" />
-* Intuitively, this is because a convolutional neural network should be able to detect features in an image no matter where they are located.<ref name="ref_d1ae" />
-* Note that the final layer of a convolutional neural network is normally fully connected.<ref name="ref_d1ae" />
-* A Convolutional Neural Network (CNN) is a deep learning algorithm that can recognize and classify features in images for computer vision.<ref name="ref_7a43">[https://missinglink.ai/guides/convolutional-neural-networks/convolutional-neural-network-architecture-forging-pathways-future/ Convolutional Neural Network Architecture: Forging Pathways to the Future]</ref>
-* The architecture of a CNN is a key factor in determining its performance and efficiency.<ref name="ref_7a43" />
-* This 7-layer CNN classified digits, digitized 32×32 pixel greyscale input images.<ref name="ref_7a43" />
-* GoogleNet (2014) Built with a CNN inspired by LetNet, the GoogleNet network, which is also named Inception V1, was made by a team at Google.<ref name="ref_7a43" />
-* LeNet was one of the very first convolutional neural networks which helped propel the field of Deep Learning.<ref name="ref_e43f">[https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/ An Intuitive Explanation of Convolutional Neural Networks]</ref>
-* The primary purpose of Convolution in case of a ConvNet is to extract features from the input image.<ref name="ref_e43f" />
-* It is important to understand that these layers are the basic building blocks of any CNN.<ref name="ref_e43f" />
-* Please note however, that these operations can be repeated any number of times in a single ConvNet.<ref name="ref_e43f" />
-* Visualizing a ConvNet trained on handwritten digits.<ref name="ref_e43f" />
-* In this post, I have tried to explain the main concepts behind Convolutional Neural Networks in simple terms.<ref name="ref_e43f" />
-* Convolutional neural networks apply a series of learnable filters to the input image.<ref name="ref_bab0">[https://codelabs.developers.google.com/codelabs/keras-flowers-convnets Convolutional neural networks, with Keras and TPUs]</ref>
-* This is how a simple convolutional neural network looks in Keras: model = tf.keras.<ref name="ref_bab0" />
-* Illustration: a convolutional neural network transforms "cubes" of data into other "cubes" of data.<ref name="ref_bab0" />
-* As you explore in-depth, computer vision and deep learning become all about the convolutional neural network (CNN).<ref name="ref_a315">[https://www.thinkautomation.com/eli5/eli5-what-is-a-convolutional-neural-network/ ELI5: what is a convolutional neural network?]</ref>
-* Convolutional neural networks were inspired by animal vision.<ref name="ref_a315" />
-* Convolutional layers are the layers that give convolutional neural networks the name.<ref name="ref_a315" />
-* In other words, pooling layers give flexibility to your convolutional neural network.<ref name="ref_a315" />
-* This is where all the features extracted by the convolutional neural network get combined.<ref name="ref_a315" />
-* Convolutional neural networks sit behind a few AI functions.<ref name="ref_a315" />
-* Convolutional neural networks let computers ‘see’ pictures.<ref name="ref_a315" />
-* Although much less common, CNNs are also being looked at to help with video analysis.<ref name="ref_a315" />
-* This overview has only scratched the surface of convolutional neural networks.<ref name="ref_a315" />
-* Abstract Deep convolutional neural networks have been successfully applied to many image-processing problems in recent works.<ref name="ref_5eff">[https://www.pnas.org/content/115/2/254 A mixed-scale dense convolutional neural network for image analysis]</ref>
-* We first introduce notation and discuss the general structure of existing deep convolutional networks.<ref name="ref_5eff" />
-* Convolutional neural networks (CNNs) model the unknown function f by using several layers that are connected to each other in succession.<ref name="ref_5eff" />
-* A schematic of a two-layer CNN architecture is shown in Fig.<ref name="ref_5eff" />
-* such that the CNN performs the task that is required.<ref name="ref_5eff" />
-* In general, the increased depth of DCNNs compared with shallow CNNs makes training more difficult.<ref name="ref_5eff" />
-* In this section, we describe a new convolutional neural network (CNN) based on the SMILES notation of compounds.<ref name="ref_5dae">[https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-018-2523-5 Convolutional neural network based on SMILES representation of compounds for detecting chemical motif]</ref>
-* An overview of our CNN is shown in Fig.<ref name="ref_5dae" />
-* Our CNN transforms the SMILES feature matrix into a low-dimensional feature vector termed the SMILES convolution finger print (SCFP).<ref name="ref_5dae" />
-* CNN has multiple layers consisting of two convolutional and pooling layers with a subsequent global pooling layer.<ref name="ref_5dae" />
-* Figure 2 shows the architecture of our CNN.<ref name="ref_5dae" />
-* Our CNN has several hyperparameters including the window size of filters, the number of filters, and others.<ref name="ref_5dae" />
-* Our CNN can be used not only as a prediction method but also as a method to compute a fingerprint.<ref name="ref_5dae" />
-* Koushik, J.: Understanding convolutional neural networks.<ref name="ref_6dee">[https://link.springer.com/article/10.1007/s13748-019-00203-0 Convolutional neural network: a review of models, methodologies and applications to object detection]</ref>
-* Gated spatio and temporal convolutional neural network for activity recognition: towards gated multimodal deep learning.<ref name="ref_6dee" />
-* A review of object detection based on convolutional neural network.<ref name="ref_6dee" />
-* Xu, H., Han, Z., Feng, S., Zhou, H., Fang, Y.: Foreign object debris material recognition based on convolutional neural networks.<ref name="ref_6dee" />
-* Rethinking Model Scaling for Convolutional Neural Networks.<ref name="ref_6dee" />
-* CNNs are not limited to image recognition, however.<ref name="ref_59a8">[https://wiki.pathmind.com/convolutional-network A Beginner's Guide to Convolutional Neural Networks (CNNs)]</ref>
-* So instead of thinking of images as two-dimensional areas, in convolutional nets they are treated as four-dimensional volumes.<ref name="ref_59a8" />
-* (Note that convolutional nets analyze images differently than RBMs.<ref name="ref_59a8" />
-* So convolutional networks perform a sort of search.<ref name="ref_59a8" />
-* The first thing to know about convolutional networks is that they don’t perceive images like humans do.<ref name="ref_59a8" />
-* Convolutional networks are designed to reduce the dimensionality of images in a variety of ways.<ref name="ref_59a8" />
-* The next layer in a convolutional network has three names: max pooling, downsampling and subsampling.<ref name="ref_59a8" />
-* Prior to CNNs, manual, time-consuming feature extraction methods were used to identify objects in images.<ref name="ref_e497">[https://www.ibm.com/cloud/learn/convolutional-neural-networks What are Convolutional Neural Networks?]</ref>
-* With each layer, the CNN increases in its complexity, identifying greater portions of the image.<ref name="ref_e497" />
-* The convolutional layer is the core building block of a CNN, and it is where the majority of computation occurs.<ref name="ref_e497" />
-* While a lot of information is lost in the pooling layer, it also has a number of benefits to the CNN.<ref name="ref_e497" />
-* Both the CNN technique and the previous pre-computed brain response atlas approach15 require a large training dataset.<ref name="ref_620c">[https://www.nature.com/articles/s41598-019-53551-1 Convolutional neural network for efficient estimation of regional brain strains]</ref>
-* This indicated some impressive generalizability and robustness of the CNN technique.<ref name="ref_620c" />
-* Further, the data-driven CNN technique does not address any physics behind brain biomechanical responses.<ref name="ref_620c" />
-* Nonetheless, the CNN can be easily re-trained to accommodate another model or a future, upgraded WHIM.<ref name="ref_620c" />
-* In recent years, CNNs have become pivotal to many computer vision applications.<ref name="ref_cf6a">[https://bdtechtalks.com/2020/01/06/convolutional-neural-networks-cnn-convnets/ What are convolutional neural networks (CNN)?]</ref>
-* The early version of CNNs, called LeNet (after LeCun), could recognize handwritten digits.<ref name="ref_cf6a" />
-* CNNs needed a lot of data and compute resources to work efficiently for large images.<ref name="ref_cf6a" />
-* Convolutional neural networks are composed of multiple layers of artificial neurons.<ref name="ref_cf6a" />
-* When you input an image into a ConvNet, each of its layers generates several activation maps.<ref name="ref_cf6a" />
-* The first (or bottom) layer of the CNN usually detects basic features such as horizontal, vertical, and diagonal edges.<ref name="ref_cf6a" />
-* The operation of multiplying pixel values by weights and summing them is called “convolution” (hence the name convolutional neural network).<ref name="ref_cf6a" />
-* A CNN is usually composed of several convolution layers, but it also contains other components.<ref name="ref_cf6a" />
-* One of the great challenges of developing CNNs is adjusting the weights of the individual neurons to extract the right features from images.<ref name="ref_cf6a" />
-* In the beginning, the CNN starts off with random weights.<ref name="ref_cf6a" />
-* The ConvNet processes each image with its random values and then compares its output with the image’s correct label.<ref name="ref_cf6a" />
-* The ConvNet goes through several epochs during training, adjusting its weights in small amounts.<ref name="ref_cf6a" />
-* As the CNN improves, the adjustments it makes to the weights become smaller and smaller.<ref name="ref_cf6a" />
-* After training the CNN, the developers use a test dataset to verify its accuracy.<ref name="ref_cf6a" />
-* Each image is run through the ConvNet, and the output is compared to the actual label of the image.<ref name="ref_cf6a" />
-* The success of convolutional neural networks is largely due to the availability of huge image datasets developed in the past decade.<ref name="ref_cf6a" />
-* You don’t, however, need to train every convolutional neural network on millions of images.<ref name="ref_cf6a" />
-* A well-trained ConvNet will tell you that it’s the image of a soldier, a child and the American flag.<ref name="ref_cf6a" />
-* These limits become more evident in practical applications of convolutional neural networks.<ref name="ref_cf6a" />
-* For instance, CNNs are now widely used to moderate content on social media networks.<ref name="ref_cf6a" />
-* Another problem with convolutional neural networks is their inability to understand the relations between different objects.<ref name="ref_cf6a" />
-* still no convolutional neural network that can solve Bongard problems with so few training examples.<ref name="ref_cf6a" />
-* In one study conducted in 2016, AI researchers trained a CNN on 20,000 Bongard samples and tested it on 10,000 more.<ref name="ref_cf6a" />
-* Today, CNNs are used in many computer vision applications such as facial recognition, image search and editing, augmented reality, and more.<ref name="ref_cf6a" />
-* Convolutional neural networks apply neural networks on images.<ref name="ref_b9b1">[https://anderfernandez.com/en/blog/how-to-create-convolutional-neural-network-keras/ How to build a convolutional neural network in Keras]</ref>
-* In this post I am going to explain what they are and how you can create a convolutional neural network in Keras with Python.<ref name="ref_b9b1" />
-* Convolutional neural networks basically take an image as input and apply different transformations that condense all the information.<ref name="ref_b9b1" />
-* A typical CNN is usually structured as a series of layers, including multiple convolutional layers and a few of fully connected layers.<ref name="ref_125c">[https://www.intechopen.com/online-first/advances-in-convolutional-neural-networks Advances in Convolutional Neural Networks]</ref>
-* Specifically, a gradient-descent based algorithm is usually adopted to iteratively optimize the parameters in a CNN.<ref name="ref_125c" />
-* Figure 1 shows the high-level abstraction of CNNs in this survey.<ref name="ref_125c" />
-* Then four methods are summarized for constructing convolutional layers in CNNs in Section 3.<ref name="ref_125c" />
-* In Section 4, we group the current CNN architectures into three types: encoder, encoder-decoder and GANs.<ref name="ref_125c" />
-* In Section 6, we give the advanced applications based on the three types of CNN structures.<ref name="ref_125c" />
-* However, it is difficult to know what size of kernels we should use in a CNN.<ref name="ref_125c" />
-* Z and F decoder represents a decoder CNN to reconstruct the input sample with Z .<ref name="ref_125c" />
-* Recall that the shortcut connection is often adopted to address the problems in deep CNNs.<ref name="ref_125c" />
-* Therefore, as shown in Figure 1, loss functions play a significant role in constructing CNNs.<ref name="ref_125c" />
-* More Specifically, if Q denotes the distribution on data, and P represents the distribution which is learned by a CNN model.<ref name="ref_125c" />
-* In this section, we summarize the typical advances that CNNs has achieved based on the three types of CNN structures.<ref name="ref_125c" />
-* Firstly, a pre-trained CNN encoder is used to extract some high-level features from an input image.<ref name="ref_125c" />
-* 6.2.4 Speech processing Note that speech signals exhibit spectral variations and correlations, CNNs are very suitable to reduce them.<ref name="ref_125c" />
-* Therefore, CNNs can also be utilized for the task of speech processing, such as speech recognition.<ref name="ref_125c" />
-* Because these two aspects are the core parts when applying CNNs into various types of tasks.<ref name="ref_125c" />
-* Convolutional networks rely on 3D architecture - height, width, and depth - to scale for image recognition.<ref name="ref_316b">[https://www.edx.org/learn/convolutional-neural-network Learn Convolutional Neural Network with Online Courses]</ref>
-* CNN Courses and Certifications Deep neural networks are critical to working with images in the era of visual data science.<ref name="ref_316b" />
-* Building your knowledge of CNN is vital to understanding image data.<ref name="ref_316b" />
-* The building blocks for true robot vision lie in convolutional neural networks where image data can fill in gaps for operation.<ref name="ref_316b" />
-* The specificity of a CNN lies in its filtering layers, which include at least one convolution layer.<ref name="ref_8bf2">[https://www.doc.ic.ac.uk/~jce317/introduction-cnns.html Convolutional Neural Networks]</ref>
-* The first major success of convolutional neural networks was AlexNet, developed by Alex Krizhevsky, in 2012 at the University of Toronto.<ref name="ref_8bf2" />
-* This current state of the art came about from ResNet, a CNN architecture from Microsoft Research.<ref name="ref_8bf2" />
-* The problem motivating ResNet was that adding extra layers to a CNN did not necessarily improve performance.<ref name="ref_8bf2" />
-* D. Wei, B. Zhou, A. Torrabla and W. Freeman, Understanding intra-class knowledge inside cnn, arXiv preprint, arXiv: 1507.02379.<ref name="ref_0379">[https://www.aimsciences.org/article/doi/10.3934/mfc.2018008 How convolutional neural networks see the world --- A survey of convolutional neural network visualization methods]</ref>
-* Visualizing and understanding convolutional networks, in Proceedings of the European Conference on Computer Vision, 2014, 818-833.<ref name="ref_0379" />
-* This paper presents a novel method for identifying coal and rock based on a deep convolutional neural network (CNN).<ref name="ref_6c40">[https://www.hindawi.com/journals/mpe/2020/2616510/ A Deep Convolutional Neural Network Model for Intelligent Discrimination between Coal and Rocks in Coal Mining Face]</ref>
-* Section 2 summarizes the basic theory of CNN.<ref name="ref_6c40" />
-* There are two stages in identifying coal and rock using convolutional neural networks: feature learning and classification.<ref name="ref_6c40" />
-* Figure 2 shows the training process of a convolutional neural network.<ref name="ref_6c40" />
-* In the application of CNN, the problem of overfitting often occurs.<ref name="ref_6c40" />
-* The training of convolutional neural network is the key to achieve the discrimination between coal and rocks.<ref name="ref_6c40" />
-* The raw data (without data augmentation) and augmented data are used to train the CNN with NET, respectively.<ref name="ref_6c40" />
-* This paper presents a method for identifying coal and rock based on a deep convolutional neural network.<ref name="ref_6c40" />
-* Some experiments are provided and the comparisons with other classical convolutional neural networks are conducted.<ref name="ref_6c40" />
-* The convolutional neural network (CNN) was first proposed in 1960s.<ref name="ref_ef11">[https://www.spiedigitallibrary.org/journals/optical-engineering/volume-58/issue-04/040901/Development-of-convolutional-neural-network-and-its-application-in-image/10.1117/1.OE.58.4.040901.full Development of convolutional neural network and its application in image classification: a survey]</ref>
-* 1 was later introduced into the research work of CNNs.<ref name="ref_ef11" />
-* The CNN model AlexNet presented by Krizhevsky et al.<ref name="ref_ef11" />
-* First, this paper introduces the history of CNN and then analyzes the development of CNN architecture in image classification.<ref name="ref_ef11" />
-* 1 , CNN architecture is generally composed of convolution layers, subsampling (pooling) layers, and fully connected layers.<ref name="ref_ef11" />
-* The combination of convolution layer, pooling layer, and fully connected layer is still the basic components of modern deep CNN.<ref name="ref_ef11" />
-* 5 AlexNet is a milestone in the development of deep CNN, which has caused a new wave of neural network research.<ref name="ref_ef11" />
-* After AlexNet achieved excellent results in the ImageNet image classification competition, researchers began to study the CNN more deeply.<ref name="ref_ef11" />
-* However, there is no clear theoretical explanation for why a CNN model can perform well.<ref name="ref_ef11" />
-* Zeiler and Fergus 31 proposed a visualization technique to understand CNNs and proposed ZFNet.<ref name="ref_ef11" />
-* 31 is to explain to a certain extent why CNNs are effective and how to improve network performance.<ref name="ref_ef11" />
-* Using small convolution kernels, VGG can make the CNN reach a depth of 19 layers.<ref name="ref_ef11" />
-* It has made a vital contribution to the development of CNNs.<ref name="ref_ef11" />
-* In addition to network initialization, the innovation of optimization method has also promoted the development of CNN.<ref name="ref_ef11" />
-* Experimental results are obtained through five similar CNN predictions.<ref name="ref_ef11" />
-* In the following, the development trend of CNNs in image classification is prospected through several aspects.<ref name="ref_ef11" />
-* A CNN has a large number of parameters, so the experiment of CNN often fails to achieve the effect of network in corresponding papers.<ref name="ref_ef11" />
-* At present, the parameter setting in training CNN is mostly based on experience and practice.<ref name="ref_ef11" />
-* Visualizing and understanding convolutional networks ,” in Eur.<ref name="ref_ef11" />
-* The final layer of the CNN architecture uses a classification layer such as softmax to provide the classification output.<ref name="ref_5ede">[https://kr.mathworks.com/discovery/convolutional-neural-network-matlab.html Convolutional Neural Network]</ref>
-* This tutorial demonstrates training a simple Convolutional Neural Network (CNN) to classify CIFAR images.<ref name="ref_1240">[https://www.tensorflow.org/tutorials/images/cnn Convolutional Neural Network (CNN)]</ref>
-* As input, a CNN takes tensors of shape (image_height, image_width, color_channels), ignoring the batch size.<ref name="ref_1240" />
-* In this example, you will configure our CNN to process inputs of shape (32, 32, 3), which is the format of CIFAR images.<ref name="ref_1240" />
-* Why ReLU is important : ReLU’s purpose is to introduce non-linearity in our ConvNet.<ref name="ref_b392">[https://medium.com/@RaghavPrabhu/understanding-of-convolutional-neural-network-cnn-deep-learning-99760835f148 Understanding of Convolutional Neural Network (CNN) — Deep Learning]</ref>
-* A ConvNet arranges its neurons in three dimensions (width, height, depth), as visualized in one of the layers.<ref name="ref_be8a">[https://cs231n.github.io/convolutional-networks/ CS231n Convolutional Neural Networks for Visual Recognition]</ref>
-* A ConvNet is made up of Layers.<ref name="ref_be8a" />
-* Together, this adds up to 290400 * 364 = 105,705,600 parameters on the first layer of the ConvNet alone.<ref name="ref_be8a" />
-* Lastly, what if we wanted to efficiently apply the original ConvNet over the image but at a stride smaller than 32 pixels?<ref name="ref_be8a" />
-* You should rarely ever have to train a ConvNet from scratch or design one from scratch.<ref name="ref_be8a" />
-* Until now we’ve omitted mentions of common hyperparameters used in each of the layers in a ConvNet.<ref name="ref_be8a" />
-* The first successful applications of Convolutional Networks were developed by Yann LeCun in 1990’s.<ref name="ref_be8a" />
-* CNN is a class of deep learning networks that has attracted much attention in recent studies.<ref name="ref_8420">[https://www.sciencedirect.com/topics/engineering/convolutional-neural-network Convolutional Neural Network - an overview]</ref>
-* The input matrix of the CNN model is an L× L×(2×D+C) matrix, where C represents the number of 2-D features.<ref name="ref_8420" />
-* The CNN model is composed of several residual blocks.<ref name="ref_8420" />
-* Finally, a three-layer CNN is employed to predict the final contact map from matrix Mat.<ref name="ref_8420" />
-* As shown, it is necessary to prepare a large number of training data with corresponding labels for efficient classification using CNN.<ref name="ref_6e8a">[https://insightsimaging.springeropen.com/articles/10.1007/s13244-018-0639-9 Convolutional neural networks: an overview and application in radiology]</ref>
-* However, one can also apply CNN to this task as well.<ref name="ref_6e8a" />
-* One way to perform segmentation is to use a CNN classifier for calculating the probability of an organ or anatomical structure.<ref name="ref_6e8a" />
-* a Denoising system with CNN in deployment phase.<ref name="ref_6e8a" />
-* By using this representation, 2D-CNN could be used to classify the reports as pulmonary embolism or not.<ref name="ref_6e8a" />
-* Their results showed that the performance of the CNN model was equivalent to or beyond that of the traditional model.<ref name="ref_6e8a" />
-* The pre-processing required in a ConvNet is much lower as compared to other classification algorithms.<ref name="ref_45f9">[https://towardsdatascience.com/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53 A Comprehensive Guide to Convolutional Neural Networks — the ELI5 way]</ref>
-* A ConvNet is able to successfully capture the Spatial and Temporal dependencies in an image through the application of relevant filters.<ref name="ref_45f9" />
-* ConvNets need not be limited to only one Convolutional Layer.<ref name="ref_45f9" />
-* CNNs are regularized versions of multilayer perceptrons.<ref name="ref_0883">[https://en.wikipedia.org/wiki/Convolutional_neural_network Convolutional neural network]</ref>
-* CNNs use relatively little pre-processing compared to other image classification algorithms.<ref name="ref_0883" />
-* The name “convolutional neural network” indicates that the network employs a mathematical operation called convolution.<ref name="ref_0883" />
-* A convolutional neural network consists of an input and an output layer, as well as multiple hidden layers.<ref name="ref_0883" />
-* The hidden layers of a CNN typically consist of a series of convolutional layers that convolve with a multiplication or other dot product.<ref name="ref_0883" />
-* Convolutional networks may include local or global pooling layers to streamline the underlying computation.<ref name="ref_0883" />
-* A distinguishing feature of CNNs is that many neurons can share the same filter.<ref name="ref_0883" />
-* The neocognitron introduced the two basic types of layers in CNNs: convolutional layers, and downsampling layers.<ref name="ref_0883" />
-* The neocognitron is the first CNN which requires units located at multiple network positions to have shared weights.<ref name="ref_0883" />
-* The first GPU-implementation of a CNN was described in 2006 by K. Chellapilla et al.<ref name="ref_0883" />
-* Subsequently, a similar GPU-based CNN by Alex Krizhevsky et al.<ref name="ref_0883" />
-* The layers of a CNN have neurons arranged in 3 dimensions: width, height and depth.<ref name="ref_0883" />
-* Distinct types of layers, both locally and completely connected, are stacked to form a CNN architecture.<ref name="ref_0883" />
-* In CNNs, each filter is replicated across the entire visual field.<ref name="ref_0883" />
-* Together, these properties allow CNNs to achieve better generalization on vision problems.<ref name="ref_0883" />
-* Another important concept of CNNs is pooling, which is a form of non-linear down-sampling.<ref name="ref_0883" />
-* This is the idea behind the use of pooling in convolutional neural networks.<ref name="ref_0883" />
-* CNNs use more hyperparameters than a standard multilayer perceptron (MLP).<ref name="ref_0883" />
-* For convolutional networks, the filter size also affects the number of parameters.<ref name="ref_0883" />
-* CNNs are often used in image recognition systems.<ref name="ref_0883" />
-* Compared to image data domains, there is relatively little work on applying CNNs to video classification.<ref name="ref_0883" />
-* However, some extensions of CNNs into the video domain have been explored.<ref name="ref_0883" />
-* CNNs have also been explored for natural language processing.<ref name="ref_0883" />
-* CNNs have been used in drug discovery.<ref name="ref_0883" />
-* CNNs have been used in the game of checkers.<ref name="ref_0883" />
-* CNNs have been used in computer Go.<ref name="ref_0883" />
-* Convolutional neural networks usually require a large amount of training data in order to avoid overfitting.<ref name="ref_0883" />
-* Therefore, they exploit the 2D structure of images, like CNNs do, and make use of pre-training like deep belief networks.<ref name="ref_0883" />
-* Central to the convolutional neural network is the convolutional layer that gives the network its name.<ref name="ref_94c6">[https://machinelearningmastery.com/convolutional-layers-for-deep-learning-neural-networks/ How Do Convolutional Layers Work in Deep Learning Neural Networks?]</ref>
-* Technically, the convolution as described in the use of convolutional neural networks is actually a “cross-correlation”.<ref name="ref_94c6" />
-===소스===
- <references />

"합성곱 신경망"의 두 판 사이의 차이

2020년 12월 15일 (화) 21:43 판

둘러보기 메뉴

검색