Successfully reported this slideshow.
Your SlideShare is downloading. ×

FIEM_ECE_Indrustrial_Training (1).pdf

Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Ad
Page | 1
Machine Learning to Deep Learning: A Journey for
Remote Sensing Data Classification
Thesis submitted in partial f...
Page | 2
< College Letter Head>
Certificate of Recommendation
This is to certify that the industrial training entitled “ M...
Page | 3
Future Institute of Engineering & Management
Affiliated to
Maulana Abul Kalam Azad University of Technology
( for...
Advertisement
Advertisement
Advertisement
Advertisement
Advertisement
Advertisement
Advertisement
Loading in …3
×

Check these out next

1 of 35 Ad
Advertisement

More Related Content

Recently uploaded (20)

Advertisement

FIEM_ECE_Indrustrial_Training (1).pdf

  1. 1. Page | 1 Machine Learning to Deep Learning: A Journey for Remote Sensing Data Classification Thesis submitted in partial fulfillment for the degree of Bachelor of Technology in Electronics and Communication Engineering of Maulana Abul Kalam Azad University of Technology ( formerly known as West Bengal University of Technology ) By Chandan Shaw (Roll No :14800319032) Under the supervision of Mr. Debabrata Pandit (Assistant Professor, Dept of ECE, FIEM) Department of Electronics & Communication Engineering Future Institute of Engineering & Management Sonarpur Station Road, Kolkata-700150 2022
  2. 2. Page | 2 < College Letter Head> Certificate of Recommendation This is to certify that the industrial training entitled “ Machine Learning to Deep Learning: A Journey for Remote Sensing Data Classification ” submitted by Chandan Shaw is absolutely based upon their own work under the supervision of Mr. Debabrata Pandit (Assistant Professor, Dept of ECE, FIEM) and that neither their thesis has been submitted for any degree/diploma or any other academic award anywhere before. ……………………………… Mr. Debabrata Pandit (Assistant Professor, Dept of ECE, FIEM) ……………………………….. Dr. Dipankar Ghosh (HOD, Dept of ECE, FIEM)
  3. 3. Page | 3 Future Institute of Engineering & Management Affiliated to Maulana Abul Kalam Azad University of Technology ( formerly West Bengal University of Technology ) Certificate of Approval* The foregoing thesis is hereby approved as a creditable study of an engineering subject carried out and presented in a manner satisfactory to warrant its acceptance as a prerequisite to the degree for which it has been submitted. It is understood that by this approval the undersigned don’t necessarily endorse or approve any statement made opinion expressed or conclusion drawn therein but approve the thesis only for the purpose for which it is submitted. Signature of the Examiners: 1……………………………………………. 2……………………………………………. 3……………………………………………. *Only in the case the thesis is approved
  4. 4. Page | 4 Acknowledgement It is a great pleasure to express my deepest gratitude and indebtedness to my (internal / external) guide(s), Mr. Debabrata Pandit Department of Electronics and Communication Engineering, Future Institute of Engineering & Management, Kolkata for his supervision, constant help and encouragement throughout the entire period. I am very much grateful to Dr. Dipankar Ghosh ,(Head of the Department), Department of Electronics and Communication Engineering, Future Institute of Engineering & Management, for necessary suggestions and helpful scientific discussions to improve the quality this thesis. I am also thankful to the all faculty members and technical staff of the department for their help whenever I needed it. Chandan Shaw Department of Electronics & Communication Engineering, Future Institute of Engineering & Management
  5. 5. Page | 5 Contents 1. …………..Abstract………………………………………………………...Page|6 2. ……...…. Introduction……………………………………………………..Page|7 3. ………….Remote Sensing………………………………………………... Page|8 4. ………….Image Resolution………………………………………………. Page|10 5. ………….Electromagnetic Spectrum……………………………………... Page|12 6. ……….…Image Classification……………………………………... Page|14 7. ……….. Data Preprocessing in Machine learning ……………………Page|15 8. …………..Python Libraries……………………………………………..… Page|23 9. …………. Fuzzy Image Classification……………………………………. Page|26 10. …………. Network Based Learning Algorithm in Deep Learning ………..Page|27 11. ……..….Conclusion……………….. ………………………………..Page|35 12. ……….....References….. ………………………………………………….Page|35
  6. 6. Page | 6 Abstract Machine learning offers the potential for effective and efficient classification of remotely sensed imagery. The strengths of machine learning include the capacity to handle data of high dimensionality and to map classes with very complex characteristics. Nevertheless, implementing a machine-learning classification is not straightforward, and the literature provides conflicting advice regarding many key issues. This article therefore provides an overview of machine learning from an applied perspective. We focus on the relatively mature methods of support vector machines, single decision trees (DTs), Random Forests, boosted DTs, artificial neural networks, and k-nearest neighbours (k-NN). Issues considered include the choice of algorithm, training data requirements, user-defined parameter selection and optimization, feature space impacts and reduction, and computational costs. We illustrate these issues through applying machine-learning classification to two publically available remotely sensed data sets. Signal and data processing has established a new standard by using deep learning (DL) and deep neural network (DNN). This is done by obtaining advanced performance in terms of audio, image, and understanding of the language naturally. The major research in remote sensing has been given to the DL (deep learning) applications. The main application of deep learning in remote sensing is the classification of the data. The quality of the data received from the remote sensing sensors is insufficient. Thus, equal importance has to be given to the challenges associated with the magnification of these low-quality images. Addressing such challenges becomes predominant as it involves different environmental conditions, the tradeoff in the imaging system, and varying altitude images. This becomes the reason for the low quality of observations, thus making classification and identification a difficult task. Another huge challenge faced in the process of classification and identification is the heterogeneous nature of the remote sensing sensors. This also affects the efficiency and effectiveness of the data from remote sensing. The approaches for processing the remote sensing data can be improved by using the multi-modal datasets from the increasing sensing and additional secondary datasets when handy applications primary consideration. This makes researchers around the world get more interest in the fusion of data from multi-source for application diversity. The temporal data integration with spectral or spatial information is made possible using the spaceborne sensor’s revisit capabilities. It helps in the representation of data structures with fresh data that is time variable, as well as data extraction techniques. Thus, this paper involves the development of the deep learning technique which is based on the remote sensing data for the extraction of features from the environmental parameters.
  7. 7. Page | 7 INTRODUCTION Remote Sensing (RS) refers to the science of identification of earth surface features and estimation of their geo-biophysical properties using electromagnetic radiation as a medium of interaction. Spectral, spatial, temporal and polarization signatures are major characteristics of the sensor/target, which facilitate target discrimination. Earth surface data as seen by the sensors in different wavelengths (reflected, scattered and/or emitted) is radiometrically and geometrically corrected before extraction of spectral information. RS data, with its ability for a synoptic view, repetitive coverage with calibrated sensors to detect changes, observations at different resolutions, provides a better alternative for natural resources management as compared to traditional methods. Indian Earth Observation (EO) programme has been applications-driven and national development has been its prime motivation. From Bhaskara to Cartosat, India's EO capability has increased manifold. Improvements are not only in spatial, spectral, temporal and radiometric resolutions, but also in their coverage and value-added products. Some of the major operational application themes, in which India has extensively used remote sensing data are agriculture, forestry, water resources, land use, urban sprawl, geology, environment, coastal zone, marine resources, snow and glacier, disaster monitoring and mitigation, infrastructure development, etc. The paper reviews RS techniques and applications carried out using both optical and microwave sensors. It also analyses the gap areas and discusses the future perspectives.
  8. 8. Page | 8 Remote Sensing: Remote sensing is a technique to observe the earth surface or the atmosphere from out ofspace using satellites (space borne) or from the air using aircrafts (airborne). Remote sensinguses a part or several parts of the electromagnetic spectrum. It records the electromagneticenergy reflected or emitted by the earth’s surface. The amount of radiation from an object(called radiance) is influenced by both the properties of the object and the radiation hittingthe object (irradiance). The human eyes register the solar light reflected by these objects andour brains interpret the colours, the grey tones and intensity variations. In remote sensingvarious kinds of tools and devices are used to make electromagnetic radiation outside thisrange from 400 to 700 nm visible to the human eye, especially the near infrared, middle-infrared, thermal-infrared and microwaves.Remote sensing imagery has many applications in mapping land-use and cover, agriculture,soils mapping, forestry, city planning, archaeological investigations, military observation, andgeomorphological surveying, land cover changes, deforestation, vegetation dynamics, waterquality dynamics, urban growth, etc. This paper starts with a brief historic overview ofremote sensing and then explains the various stages and the basic principles of remotelysensed data collection mechanism. Remote sensing is a technique of obtaining informationabout objects through the analysis of data collected by special instruments that are not inphysical contact with the objects of investigation. From a general perspectives, remotesensing is the science of acquiring and analyzing information about objects or phenomenafrom a distance (Jensen, 2000, Lilles and and Keifer, 1987). However, conventionally, remotesensing (RS) refers to the identification of earth features by detecting the characteristicselectromagnetic radiation that is reflected/emitted by the earth surface. The sensors on-boardvarious platforms detect the radiation received from the targets in different spectral regions. Sensors Types: Remote sensing uses a sensor to capture an image. For example, airplanes, satellites, and UAVs have specialized platforms that carry sensors. The diagram below shows the major remote sensing technologies and their typical altitudes. Types of Remote Sensing: The two types of remote sensing sensors are:  Passive sensors  Active sensors Active Sensor: The main difference between active sensors is that this type of sensor illuminates its target. Then, active sensors measure the reflected light. For example, Radarsat-2 is an active sensor that uses synthetic aperture radar.
  9. 9. Page | 9 Imagine the flash of a camera. It brightens its target. Next, it captures the return light. This is the same principle of how active sensors work. Passive Sensor: Passive sensors measure reflected light emitted from the sun. When sunlight reflects off Earth’s surface, passive sensors capture that light. For example, Landsat and Sentinel are passive sensors. They capture images by sensing reflected sunlight in the electromagnetic spectrum. Passive remote sensing measures reflected energy emitted from the sun. Whereas active remote sensing illuminates its target and measures its backscatter.
  10. 10. Page | 10 Image Resolution: For earth observation, you also have to consider image resolution. Remote sensing divides image resolution into three different types:  Spatial resolution  Spectral resolution  Temporal resolution Spatial resolution: Spatial resolution is the detail in pixels of an image. High spatial resolution means more detail and smaller pixel size. Whereas, lower spatial resolution means less detail and larger pixel size. Typically, drones like DJI capture images with one of the highest spatial resolution. Even though satellites are highest in the atmosphere, they are capable of 50cm pixel size or greater.
  11. 11. Page | 11 Spectral resolution: Spectral Resolution is the amount of spectral detail in a band. High spectral resolution means its bands are more narrow. Whereas low spectral resolution has broader bands covering more of the spectrum. Temporal resolution: Temporal Resolution is the time it takes for a satellite to complete a full orbit. UAVs, airplanes, and helicopters are completely flexible. But satellites orbit the Earth in set paths. Global position system satellites are in medium Earth orbit (MEO). Because they follow a continuous orbital path, revisit times are consistent. This means our GPS receiver can almost always achieve 3 satellites or greater for high accuracy.
  12. 12. Page | 12 The Electromagnetic Spectrum: The electromagnetic spectrum ranges from short wavelengths (like X-rays) to long wavelengths (like radio waves). Our eyes only see the visible range (red, green, and blue). But other types of sensors can see beyond human vision. Ultimately, this is why remote sensing is so powerful. Our eyes are sensitive to the visible spectrum (390-700 nm). But engineers design sensors to capture beyond these wavelengths in the atmospheric window. For example, near-infrared (NIR) is in the 700-1400 nm range. Vegetation reflects more green light because that’s how our eyes see it. But it’s even more sensitive to near-infrared. That’s why we use indexes like NDVI to classify vegetation.
  13. 13. Page | 13 Spectral Bands: Spectral bands are groups of wavelengths. For example, ultraviolet, visible, near-infrared, thermal infrared, and microwave are spectral bands. We categorize each spectral region based on its frequency (v) or wavelength. There are two types of imagery for passive sensors:  Multispectral imagery  Hyperspectral imagery The main difference between multispectral and hyperspectral is the number of bands and how narrow the bands are. Hyperspectral images have hundreds of narrow bands, multispectral images consist of 3-10 wider bands.  Multispectral: Multispectral imagery generally refers to 3 to 10 bands. For example, Landsat-8 produces 11 separate images for each scene.  Coastal aerosol (0.43-0.45 um)  Blue (0.45-0.51 um)  Green (0.53-0.59 um)  Red (0.64-0.67 um)  Near infrared NIR (0.85-0.88 um)  Short-wave infrared SWIR 1 (1.57-1.65 um)  Short-wave infrared SWIR 2 (2.11-2.29 um)  Panchromatic (0.50-0.68 um)  Cirrus (1.36-1.38 um)  Thermal infrared TIRS 1 (10.60-11.19 um)  Thermal infrared TIRS 2 (11.50-12.51 um)
  14. 14. Page | 14  Hyperspectral: Hyperspectral imagery has much narrower bands (10-20 nm). A hyperspectral image has hundreds of thousands of bands. For example, Hyperion (part of the EO-1 satellite) produces 220 spectral bands (0.4-2.5 um). Image Classification: When you examine a photo and you try to pull out features and characteristics from it, this is the act of using image interpretation. We use image interpretation in forestry, military, and urban environments. We can interpret features because all objects have their own unique chemical composition. In remote sensing, we distinguish these differences by obtaining their spectral signature. When you assign classes to features on the ground, this is the process of image classification. The three main methods to classify images are:  Supervised classification  Unsupervised classification  Object-based image analysis
  15. 15. Page | 15 The goal of image classification is to produce land use/land cover. By using remote sensing software, this is how we classify water, wetlands, trees, and urban areas in land cover. Data Preprocessing in Machine learning: Data preprocessing is a process of preparing the raw data and making it suitable for a machine learning model. It is the first and crucial step while creating a machine learning model. When creating a machine learning project, it is not always a case that we come across the clean and formatted data. And while doing any operation with data, it is mandatory to clean it and put in a formatted way. So for this, we use data preprocessing task. Need Data Preprocessing: A real-world data generally contains noises, missing values, and maybe in an unusable format which cannot be directly used for machine learning models. Data preprocessing is required tasks for cleaning the data and making it suitable for a machine learning model which also increases the accuracy and efficiency of a machine learning model. It involves below steps: o Getting the dataset o Importing libraries o Importing datasets
  16. 16. Page | 16 o Finding Missing Data o Encoding Categorical Data o Splitting dataset into training and test set o Feature scaling Supervised Machine Learning: Supervised learning is the types of machine learning in which machines are trained using well "labelled" training data, and on basis of that data, machines predict the output. The labelled data means some input data is already tagged with the correct output. In supervised learning, the training data provided to the machines work as the supervisor that teaches the machines to predict the output correctly. It applies the same concept as a student learns in the supervision of the teacher. Supervised learning is a process of providing input data as well as correct output data to the machine learning model. The aim of a supervised learning algorithm is to find a mapping function to map the input variable(x) with the output variable(y). In the real-world, supervised learning can be used for Risk Assessment, Image classification, Fraud Detection, spam filtering, etc. How Supervised Learning Works? In supervised learning, models are trained using labelled dataset, where the model learns about each type of data. Once the training process is completed, the model is tested on the basis of test data (a subset of the training set), and then it predicts the output. The working of Supervised learning can be easily understood by the below example and diagram:
  17. 17. Page | 17 Suppose we have a dataset of different types of shapes which includes square, rectangle, triangle, and Polygon. Now the first step is that we need to train the model for each shape. o If the given shape has four sides, and all the sides are equal, then it will be labelled as a Square. o If the given shape has three sides, then it will be labelled as a triangle. o If the given shape has six equal sides then it will be labelled as hexagon. Now, after training, we test our model using the test set, and the task of the model is to identify the shape. The machine is already trained on all types of shapes, and when it finds a new shape, it classifies the shape on the bases of a number of sides, and predicts the output.
  18. 18. Page | 18 Types of supervised Machine learning Algorithms: Supervised learning can be further divided into two types of problems:
  19. 19. Page | 19 1. Regression Regression algorithms are used if there is a relationship between the input variable and the output variable. It is used for the prediction of continuous variables, such as Weather forecasting, Market Trends, etc. Below are some popular Regression algorithms which come under supervised learning: o Linear Regression o Regression Trees o Non-Linear Regression o Bayesian Linear Regression o Polynomial Regression 2. Classification Classification algorithms are used when the output variable is categorical, which means there are two classes such as Yes-No, Male-Female, True-false, etc. Spam Filtering, o Random Forest o Decision Trees o Logistic Regression o Support vector Machines Advantages of Supervised learning: o With the help of supervised learning, the model can predict the output on the basis of prior experiences. o In supervised learning, we can have an exact idea about the classes of objects. o Supervised learning model helps us to solve various real-world problems such as fraud detection, spam filtering, etc.
  20. 20. Page | 20 Disadvantages of supervised learning: o Supervised learning models are not suitable for handling the complex tasks. o Supervised learning cannot predict the correct output if the test data is different from the training dataset. o Training required lots of computation times. o In supervised learning, we need enough knowledge about the classes of object. Unsupervised Machine Learning In the previous topic, we learned supervised machine learning in which models are trained using labeled data under the supervision of training data. But there may be many cases in which we do not have labeled data and need to find the hidden patterns from the given dataset. So, to solve such types of cases in machine learning, we need unsupervised learning techniques. What is Unsupervised Learning? As the name suggests, unsupervised learning is a machine learning technique in which models are not supervised using training dataset. Instead, models itself find the hidden patterns and insights from the given data. It can be compared to learning which takes place in the human brain while learning new things. It can be defined as: “Unsupervised learning is a type of machine learning in which models are trained using unlabeled dataset and are allowed to act on that data without any supervision.” Unsupervised learning cannot be directly applied to a regression or classification problem because unlike supervised learning, we have the input data but no corresponding output data. The goal of unsupervised learning is to find the underlying structure of dataset, group that data according to similarities, and represent that dataset in a compressed format. Example: Suppose the unsupervised learning algorithm is given an input dataset containing images of different types of cats and dogs. The algorithm is never trained upon the given dataset, which means it does not have any idea about the features of the dataset. The task of the unsupervised learning algorithm is to identify the image features on their own. Unsupervised learning algorithm will perform this task by clustering the image dataset into the groups according to similarities between images.
  21. 21. Page | 21 Why use Unsupervised Learning? Below are some main reasons which describe the importance of Unsupervised Learning: o Unsupervised learning is helpful for finding useful insights from the data. o Unsupervised learning is much similar as a human learns to think by their own experiences, which makes it closer to the real AI. o Unsupervised learning works on unlabeled and uncategorized data which make unsupervised learning more important. o In real-world, we do not always have input data with the corresponding output so to solve such cases, we need unsupervised learning. Working of Unsupervised Learning Working of unsupervised learning can be understood by the below diagram:
  22. 22. Page | 22 Here, we have taken an unlabeled input data, which means it is not categorized and corresponding outputs are also not given. Now, this unlabeled input data is fed to the machine learning model in order to train it. Firstly, it will interpret the raw data to find the hidden patterns from the data and then will apply suitable algorithms such as k- means clustering, Decision tree, etc. Once it applies the suitable algorithm, the algorithm divides the data objects into groups according to the similarities and difference between the objects.
  23. 23. Page | 23 Python libraries that are used in Machine Learning are:  Numpy  Scipy  Scikit-learn  Theano  TensorFlow  Keras  PyTorch  Pandas  Matplotlib TensorFlow: TensorFlow is a very popular open-source library for high performance numerical computation developed by the Google Brain team in Google. As the name suggests, Tensorflow is a framework that involves defining and running computations involving tensors. It can train and run deep neural networks that can be used to develop several AI applications. TensorFlow is widely used in the field of deep learning research and application. # Python program using TensorFlow # for multiplying two arrays # import `tensorflow` import tensorflow as tf # Initialize two constants x1 = tf.constant([1, 2, 3, 4]) x2 = tf.constant([5, 6, 7, 8]) # Multiply result = tf.multiply(x1, x2) # Initialize the Session sess = tf.Session() # Print the result print(sess.run(result)) # Close the session sess.close() Output: [ 5 12 21 32]
  24. 24. Page | 24 Use of TensorFlow in Deep Learning:  Handling deep neural networks  Natural Language Processing  Partial Differential Equation  TensorFlow  Abstraction capabilities  Image, Text, and Speech recognition  Effortless collaboration of ideas and code Keras: It provides many inbuilt methods for groping, combining and filtering data. Keras is a very popular Machine Learning library for Python. It is a high-level neural networks API capable of running on top of TensorFlow, CNTK, or Theano. It can run seamlessly on both CPU and GPU. Keras makes it really for ML beginners to build and design a Neural Network. One of the best thing about Keras is that it allows for easy and fast prototyping. Use of Keras in Deep Learning:  Neural layers  Activation and cost functions  Batch normalization  Dropout  Pooling Linear Regression: import numpy as np from sklearn.linear_model import LinearRegression X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]]) #y=1*x_0+2*x_1+3 y = np.dot(X, np.array([1, 2])) + 3+np.random.rand() reg = LinearRegression() reg.fit(x, y)
  25. 25. Page | 25 reg.score(x, y) reg.coef reg.intercept_ reg.predict(np.array ([[3, 5]]) ) Crop Yield Estimation: import pandas. from sklearn import linear_model df=pandas.read_excel("/content/drive/MyDrive/MLData/CropYeildPredictionSampleData.xlsx") df-df.dropna(how="any") df= df.drop(columns=[District."Year". Crop', 'Production(Tonnes)]) y=df.pop("Yield(Tonnes/Ha)") x=df.values yay.to_numpy() y= y.reshape(-1.1) model yield linear model. LinearRegression(normalize=True) model yield.fit(x, y) model_yield.predict(x[0].reshape(1,-1)) pred_y-regr.predict(x)
  26. 26. Page | 26 Fuzzy Image Classification: In traditional classification methods such as minimum distance method, each pixel or each segment in the image will have an attribute equal to 1 or 0 expressing whether the pixel or segment belongs to a certain class or not, respectively. In fuzzy classification, instead of a binary decision-making, the possibility of each pixel/segment belonging to a specific class is considered, which is defined using membership functions. A membership function offers membership degree values ranging from 0 to 1, where 1 means fully belonging to the class and 0 means not belonging to the class [24]. Implementing fuzzy logic ensures that the borders are not crisp thresholds any more, but membership functions within which each parameter value will have a specific probability to be assigned to a specific class are used. Appending more parameters to this classification, for example, using NIR ratio and NDVI for vegetation classification, better results will be achieved. Using fuzzy logic, classification accuracy is less sensitive to the thresholds. is a fuzzy membership function over domain X. is called membership degree, which ranges from 0 to 1 over domain X [23,24]. can be a Gaussian, Triangular, Trapezoidal, or other standard functions depending on the application. In this research, trapezoidal and triangular functions are used. Neural Network: A neural network is structured like the human brain and consists of artificial neurons, also known as nodes. These nodes are stacked next to each other in three layers:  The input layer  The hidden layer(s)  The output layer Data provides each node with information in the form of inputs. The node multiplies the inputs with random weights, calculates them, and adds a bias. Finally, nonlinear functions, also known as activation functions, are applied to determine which neuron to fire.
  27. 27. Page | 27 Network Based Learning Algorithm: 1. Convolutional Neural Networks (CNNs) 2. Long Short Term Memory Networks (LSTMs) 3. Recurrent Neural Networks (RNNs) 4. Generative Adversarial Networks (GANs) 5. Radial Basis Function Networks (RBFNs) 1. Convolutional Neural Networks (CNNs): It is also known as ConvNets, consist of multiple layers and are mainly used for image processing and object detection. Yann LeCun developed the first CNN in 1988 when it was called LeNet. It was used for recognizing characters like ZIP codes and digits. CNN's are widely used to identify satellite images, process medical images, forecast time series, and detect anomalies. How Do CNNs Work? CNN's have multiple layers that process and extract features from data: Convolution Layer  CNN has a convolution layer that has several filters to perform the convolution operation. Rectified Linear Unit (ReLU)  CNN's have a ReLU layer to perform operations on elements. The output is a rectified feature map.
  28. 28. Page | 28 Pooling Layer  The rectified feature map next feeds into a pooling layer. Pooling is a down-sampling operation that reduces the dimensions of the feature map.  The pooling layer then converts the resulting two-dimensional arrays from the pooled feature map into a single, long, continuous, linear vector by flattening it. Fully Connected Layer  A fully connected layer forms when the flattened matrix from the pooling layer is fed as an input, which classifies and identifies the images. Below is an example of an image processed via CNN. 2. Long Short Term Memory Network: LSTMs are a type of Recurrent Neural Network (RNN) that can learn and memorize long-term dependencies. Recalling past information for long periods is the default behavior. LSTMs retain information over time. They are useful in time-series prediction because they remember previous inputs. LSTMs have a chain-like structure where four interacting layers communicate in a unique way. Besides time-series predictions, LSTMs are typically used for speech recognition, music composition, and pharmaceutical development.
  29. 29. Page | 29 How Do LSTMs Work?  First, they forget irrelevant parts of the previous state  Next, they selectively update the cell-state values  Finally, the output of certain parts of the cell state Below is a diagram of how LSTMs operate: 3. Recurrent Neural Networks(RNNs): RNNs have connections that form directed cycles, which allow the outputs from the LSTM to be fed as inputs to the current phase. The output from the LSTM becomes an input to the current phase and can memorize previous inputs due to its internal memory. RNNs are commonly used for image captioning, time-series analysis, natural-language processing, handwriting recognition, and machine translation. An unfolded RNN looks like this:
  30. 30. Page | 30 How Do RNNs work?  The output at time t-1 feeds into the input at time t.  Similarly, the output at time t feeds into the input at time t+1.  RNNs can process inputs of any length.  The computation accounts for historical information, and the model size does not increase with the input size. Here is an example of how Google’s autocompleting feature works:
  31. 31. Page | 31 4. Generative Adversarial Networks (GANs): GANs are generative deep learning algorithms that create new data instances that resemble the training data. GAN has two components: a generator, which learns to generate fake data, and a discriminator, which learns from that false information. The usage of GANs has increased over a period of time. They can be used to improve astronomical images and simulate gravitational lensing for dark-matter research. Video game developers use GANs to upscale low-resolution, 2D textures in old video games by recreating them in 4K or higher resolutions via image training. GANs help generate realistic images and cartoon characters, create photographs of human faces, and render 3D objects. How Do GANs work?  The discriminator learns to distinguish between the generator’s fake data and the real sample data.  During the initial training, the generator produces fake data, and the discriminator quickly learns to tell that it's false.  The GAN sends the results to the generator and the discriminator to update the model. Below is a diagram of how GANs operate:
  32. 32. Page | 32 5. Radial Basic Function Networks (RBFNs): RBFNs are special types of feedforward neural networks that use radial basis functions as activation functions. They have an input layer, a hidden layer, and an output layer and are mostly used for classification, regression, and time-series prediction. How Do RBFNs Work?  RBFNs perform classification by measuring the input's similarity to examples from the training set.  RBFNs have an input vector that feeds to the input layer. They have a layer of RBF neurons.  The function finds the weighted sum of the inputs, and the output layer has one node per category or class of data.  The neurons in the hidden layer contain the Gaussian transfer functions, which have outputs that are inversely proportional to the distance from the neuron's center.  The network's output is a linear combination of the input’s radial-basis functions and the neuron’s parameters. See this example of an RBFN:
  33. 33. Page | 33 Deep Learning concepts through RNN, R-CNN, Faster RCNN, SSD, YOLO etc & their applications: R-CNN: Object detection is the process of finding and classifying objects in an image. One deep learning approach, regions with convolutional neural networks (R-CNN), combines rectangular region proposals with convolutional neural network features. R-CNN is a two-stage detection algorithm. The first stage identifies a subset of regions in an image that might contain an object. The second stage classifies the object in each region. Applications for R-CNN object detectors include:  Autonomous driving  Smart surveillance systems  Facial recognition Computer Vision Toolbox™ provides object detectors for the R-CNN, Fast R-CNN, and Faster R-CNN algorithms. Faster R-CNN The Faster R-CNN[4] detector adds a region proposal network (RPN) to generate region proposals directly in the network instead of using an external algorithm like Edge Boxes. The RPN uses Anchor Boxes for Object Detection. Generating region proposals in the network is faster and better tuned to your data. Use the trainFasterRCNNObjectDetector function to train a Faster R-CNN object detector. The function returns a fasterRCNNObjectDetector that detects objects from an image. Yolo: All of the previous object detection algorithms use regions to localize the object within the image. The network does not look at the complete image. Instead, parts of the image which have high probabilities of containing the object. YOLO or You Only Look Once is an object detection algorithm much different from the region based algorithms seen above. In YOLO a single convolutional network predicts the bounding boxes and the class probabilities for these boxes.
  34. 34. Page | 34 YOLO How YOLO works is that we take an image and split it into an SxS grid, within each of the grid we take m bounding boxes. For each of the bounding box, the network outputs a class probability and offset values for the bounding box. The bounding boxes having the class probability above a threshold value is selected and used to locate the object within the image. YOLO is orders of magnitude faster(45 frames per second) than other object detection algorithms. The limitation of YOLO algorithm is that it struggles with small objects within the image, for example it might have difficulties in detecting a flock of birds. This is due to the spatial constraints of the algorithm.
  35. 35. Page | 35 Conclusion: We have discussed the main areas where ML can make a major impact in geosciences and remote sensing. ML focuses on the automatically extraction of information from data by computational and statistical methods. Herein, the features of the ML techniques for nonparametric regression and classification purposes are outlined. The ML's application areas are very diverse and include different themes such as trace gases, aerosol products, vegetation indices, ocean products, characterization of rock mass, liquefaction phenomenon, ground motion parameters, interpreting the remote sensing image, etc. We also presented a review of a number of recent applications of the new GP method in the field. Two illustrative examples are presented to demonstrate the efficiency of ML for tackling the geosciences and remote sensing problems. Currently, data analysis methods play a central role in geosciences and remote sensing. While gathering large collections of data is essential in the field, analyzing this information becomes more challenging. Evidently, such “Big Data” has notable effects both on the predictive analytics and the knowledge extraction and interpretation tools. Considering the significant capabilities of ML, it seems to be a very efficacious approach to handle this type of information. References: 1. Arnell, N.W.; Gosling, S.N. The impacts of climate change on river flood risk at the global scale. Clim. Chang. 2016, 134, 387–401. [CrossRef] 2. Garg, A., Garg, Ankit, Tai, K., Sreedeep, S., 2014a. An integrated SRM-multi- gene genetic programming approach for prediction of factor of safety of 3-D soil nailed slopes. Engineering Applications of Artificial Intelligence 30, 30e40. 3. Carpenter, G.A., Gjaja, M.N., Gopal, S., Woodcock, C.E., 1997. Art neural networks for remote sensing: vegetation classification from landsat tm and terrain data. IEEE Transactions on Geoscience and Remote Sensing 35 (2), 308e325. 4. Brown, M.E., Lary, D.J., Vrieling, A., Stathakis, D., Mussa, H., 2008. Neural networks as a tool for constructing continuous NDVI time series from AVHRR and MODIS. International Journal of Remote Sensing 29 (24), 7141e7158

×