Search results for: ground truth for supervised learning

Authors: Mais Nijim, Rama Devi Chennuboyina, Waseem Al Aqqad

Abstract:

Advances in spatial and spectral resolution of satellite images have led to tremendous growth in large image databases. The data we acquire through satellites, radars and sensors consists of important geographical information that can be used for remote sensing applications such as region planning, disaster management. Spatial data classification and object recognition are important tasks for many applications. However, classifying objects and identifying them manually from images is a difficult task. Object recognition is often considered as a classification problem, this task can be performed using machine-learning techniques. Despite of many machine-learning algorithms, the classification is done using supervised classifiers such as Support Vector Machines (SVM) as the area of interest is known. We proposed a classification method, which considers neighboring pixels in a region for feature extraction and it evaluates classifications precisely according to neighboring classes for semantic interpretation of region of interest (ROI). A dataset has been created for training and testing purpose; we generated the attributes by considering pixel intensity values and mean values of reflectance. We demonstrated the benefits of using knowledge discovery and data-mining techniques, which can be on image data for accurate information extraction and classification from high spatial resolution remote sensing imagery.

Keywords: remote sensing, object recognition, classification, data mining, waterbody identification, feature extraction

Procedia PDF Downloads 338

Authors: Akeel A. Shah, Tong Zhang

Abstract:

Computational approaches to learning the properties of materials are commonplace, motivated by the need to screen or design materials for a given application, e.g., semiconductors and energy storage. Experimental approaches can be both time consuming and costly. Unfortunately, computational approaches such as ab-initio electronic structure calculations and classical or ab-initio molecular dynamics are themselves can be too slow for the rapid evaluation of materials, often involving thousands to hundreds of thousands of candidates. Machine learning assisted approaches have been developed to overcome the time limitations of purely physics-based approaches. These approaches, on the other hand, require large volumes of data for training (hundreds of thousands on many standard data sets such as QM7b). This means that they are limited by how quickly such a large data set of physics-based simulations can be established. At high fidelity, such as configuration interaction, composite methods such as G4, and coupled cluster theory, gathering such a large data set can become infeasible, which can compromise the accuracy of the predictions - many applications require high accuracy, for example band structures and energy levels in semiconductor materials and the energetics of charge transfer in energy storage materials. In order to circumvent this problem, multi-fidelity approaches can be adopted, for example the Δ-ML method, which learns a high-fidelity output from a low-fidelity result such as Hartree-Fock or density functional theory (DFT). The general strategy is to learn a map between the low and high fidelity outputs, so that the high-fidelity output is obtained a simple sum of the physics-based low-fidelity and correction, Although this requires a low-fidelity calculation, it typically requires far fewer high-fidelity results to learn the correction map, and furthermore, the low-fidelity result, such as Hartree-Fock or semi-empirical ZINDO, is typically quick to obtain, For high-fidelity outputs the result can be an order of magnitude or more in speed up. In this work, a new multi-fidelity approach is developed, based on a graph convolutional network (GCN) combined with semi-supervised learning. The GCN allows for the material or molecule to be represented as a graph, which is known to improve accuracy, for example SchNet and MEGNET. The graph incorporates information regarding the numbers of, types and properties of atoms; the types of bonds; and bond angles. They key to the accuracy in multi-fidelity methods, however, is the incorporation of low-fidelity output to learn the high-fidelity equivalent, in this case by learning their difference. Semi-supervised learning is employed to allow for different numbers of low and high-fidelity training points, by using an additional GCN-based low-fidelity map to predict high fidelity outputs. It is shown on 4 different data sets that a significant (at least one order of magnitude) increase in accuracy is obtained, using one to two orders of magnitude fewer low and high fidelity training points. One of the data sets is developed in this work, pertaining to 1000 simulations of quinone molecules (up to 24 atoms) at 5 different levels of fidelity, furnishing the energy, dipole moment and HOMO/LUMO.

Keywords: .materials screening, computational materials, machine learning, multi-fidelity, graph convolutional network, semi-supervised learning

Procedia PDF Downloads 39

Authors: Farid Khosravikia, Patricia Clayton

Abstract:

The main objective of this study is to evaluate the advantages and disadvantages of various machine learning techniques in forecasting ground-motion intensity measures given source characteristics, source-to-site distance, and local site condition. Intensity measures such as peak ground acceleration and velocity (PGA and PGV, respectively) as well as 5% damped elastic pseudospectral accelerations at different periods (PSA), are indicators of the strength of shaking at the ground surface. Estimating these variables for future earthquake events is a key step in seismic hazard assessment and potentially subsequent risk assessment of different types of structures. Typically, linear regression-based models, with pre-defined equations and coefficients, are used in ground motion prediction. However, due to the restrictions of the linear regression methods, such models may not capture more complex nonlinear behaviors that exist in the data. Thus, this study comparatively investigates potential benefits from employing other machine learning techniques as a statistical method in ground motion prediction such as Artificial Neural Network, Random Forest, and Support Vector Machine. The algorithms are adjusted to quantify event-to-event and site-to-site variability of the ground motions by implementing them as random effects in the proposed models to reduce the aleatory uncertainty. All the algorithms are trained using a selected database of 4,528 ground-motions, including 376 seismic events with magnitude 3 to 5.8, recorded over the hypocentral distance range of 4 to 500 km in Oklahoma, Kansas, and Texas since 2005. The main reason of the considered database stems from the recent increase in the seismicity rate of these states attributed to petroleum production and wastewater disposal activities, which necessities further investigation in the ground motion models developed for these states. Accuracy of the models in predicting intensity measures, generalization capability of the models for future data, as well as usability of the models are discussed in the evaluation process. The results indicate the algorithms satisfy some physically sound characteristics such as magnitude scaling distance dependency without requiring pre-defined equations or coefficients. Moreover, it is shown that, when sufficient data is available, all the alternative algorithms tend to provide more accurate estimates compared to the conventional linear regression-based method, and particularly, Random Forest outperforms the other algorithms. However, the conventional method is a better tool when limited data is available.

Keywords: artificial neural network, ground-motion models, machine learning, random forest, support vector machine

Procedia PDF Downloads 122

Authors: Hanene Sahli, Aymen Mouelhi, Marwa Hajji, Amine Ben Slama, Mounir Sayadi, Farhat Fnaiech, Radhwane Rachdi

Abstract:

Fetal abnormality is still a public health problem of interest to both mother and baby. Head defect is one of the most high-risk fetal deformities. Fetal head categorization is a sensitive task that needs a massive attention from neurological experts. In this sense, biometrical measurements can be extracted by gynecologist doctors and compared with ground truth charts to identify normal or abnormal growth. The fetal head biometric measurements such as Biparietal Diameter (BPD), Occipito-Frontal Diameter (OFD) and Head Circumference (HC) needs to be monitored, and expert should carry out its manual delineations. This work proposes a new approach to automatically compute BPD, OFD and HC based on morphological characteristics extracted from head shape. Hence, the studied data selected at the same Gestational Age (GA) from the fetal Ultrasound images (US) are classified into two categories: Normal and abnormal. The abnormal subjects include hydrocephalus, microcephaly and dolichocephaly anomalies. By the use of a support vector machines (SVM) method, this study achieved high classification for automated detection of anomalies. The proposed method is promising although it doesn't need expert interventions.

Keywords: biometric measurements, fetal head malformations, machine learning methods, US images

Procedia PDF Downloads 288

Authors: Anum Ali, Kashaf ad Dooja, Asif Saleem

Abstract:

The future smart cities trend will be towards Internet of Things (IoT); IoT creates dynamic connections in a ubiquitous manner. Smart cities offer ease and flexibility for daily life matters. By using small devices that are connected to cloud servers based on IoT, network traffic between these devices is growing exponentially, whose security is a concerned issue, since ratio of cyber attack may make the network traffic vulnerable. This paper discusses the latest machine learning approaches in related work further to tackle the increasing rate of cyber attacks, machine learning algorithm is applied to IoT-based network traffic data. The proposed algorithm train itself on data and identify different sections of devices interaction by using supervised learning which is considered as a classifier related to a specific IoT device class. The simulation results clearly identify the attacks and produce fewer false detections.

Keywords: IoT, traffic security, deep learning, classification

Procedia PDF Downloads 152

Authors: Haleh Dadgostar, Sara Kaviani, Hanieh Adib, Ali Mazaherinezhad, Masoud Solaymani-Dodaran, Fahimeh Soheilipour, Abdolreza Pazouki

Abstract:

Background and Objectives: Effectiveness of various exercise protocols in weight reduction after bariatric surgery has not been sufficiently explored in the literature. We compared the effect of minimally supervised home-based and closely supervised Gym based exercise programs on weight reduction and insulin resistance after bariatric surgery. Methods: Women undergoing gastric bypass surgery were invited to participate in an exercise program and were randomly allocated into two groups. They were either offered a minimally supervised home-based (MSHB) or closely supervised Gym-based (CSGB) exercise program. The CSGB protocol constitute two sessions per week of training under ACSM guidelines. In the MSHB protocol participants received a notebook containing a list of recommended aerobic and resistance exercises, a log to record their activity and a schedule of follow up phone calls and clinic visits. Both groups received a pedometer. We measured their weight, BMI, lipid profile, FBS, and insulin level at the baseline and after 20 weeks of exercise and were compared at the end of the study. Results: A total of 80 patients completed our study (MSHB=38 and CSGB=42). The baseline comparison showed that the two groups are similar. Using the ANCOVA method of analysis the mean change in BMI (covariate: BMI at the beginning of the study) was slightly better in CSGB compared with the MSHB (between-group mean difference: 3.33 (95%CI 4.718 to 1.943, F: 22.844 p < 0.001)). Conclusion: Our results showed that both MSHB and CSGB exercise methods are somewhat equally effective in improvement of studied factors in the two groups. With considerably lower costs of Minimally Supervised Home Based exercise programs, these methods should be considered when adequate funding are not available.

Keywords: postoperative exercise, insulin resistance, bariatric surgery, morbid obesity

Procedia PDF Downloads 288

Authors: Khalil Khan, Ikram Syed, Muhammad Ehsan Mazhar, Iran Uddin, Nasir Ahmad

Abstract:

Face parts labeling is the process of assigning class labels to each face part. A face parts labeling method (FPL) which divides a given image into its constitutes parts is proposed in this paper. A database FaceD consisting of 564 images is labeled with hand and make publically available. A supervised learning model is built through extraction of features from the training data. The testing phase is performed with two semantic segmentation methods, i.e., pixel and super-pixel based segmentation. In pixel-based segmentation class label is provided to each pixel individually. In super-pixel based method class label is assigned to super-pixel only – as a result, the same class label is given to all pixels inside a super-pixel. Pixel labeling accuracy reported with pixel and super-pixel based methods is 97.68 % and 93.45% respectively.

Keywords: face labeling, semantic segmentation, classification, face segmentation

Procedia PDF Downloads 255

Authors: Ágoston Nagy

Abstract:

This paper introduces a method of using unsupervised machine learning to visualize the feature space of a dataset in 2D, in order to find most characteristic segments in the set. After dimension reduction, users can select clusters by manual drawing. Selected clusters are recorded into a data model that is used for later predictions, based on realtime data. Predictions are made with supervised learning, using Gesture Recognition Toolkit. The paper introduces two example applications: a semantic audio organizer for analyzing incoming sounds, and a gesture database organizer where gestural data (recorded by a Leap motion) is visualized for further manipulation.

Keywords: gesture recognition, machine learning, real-time interaction, visualization

Procedia PDF Downloads 353

Authors: Peter Antwi Buah, Yingbin Zhang, Jianxian He, Chenlin Xiang, Delali Atsu Y. Bakah

Abstract:

Near-fault ground motions with velocity pulses are considered to cause significant damage to structures or slopes compared to ordinary ground motions without velocity pulses. The double pulsed pulse-like ground motion is as well known to be stronger than the single pulse. This study has numerically justified this perspective by studying the dynamic response of a homogeneous rock slope subjected to four pulse-like and two non-pulse-like ground motions using the Fast Lagrangian Analysis of Continua in 3 Dimensions (FLAC3D) software. Two of the pulse-like ground motions just have a single pulse. The results show that near-fault ground motions with velocity pulses can cause a higher dynamic response than regular ground motions. The amplification of the peak ground acceleration (PGA) in horizontal direction increases with the increase of the slope elevation. The seismic response of the slope under double pulse ground motion is stronger than that of the single pulse ground motion. The PGV amplification factor under the effect of the non-pulse-like records is also smaller than those under the pulse-like records. The velocity pulse strengthens the earthquake damage to the slope, which results in producing a more strong dynamic response.

Keywords: velocity pulses, dynamic response, PGV magnification effect, elevation effect, double pulse

Procedia PDF Downloads 176

Authors: Said Alshukri, Mazhar Hussain Malik

Abstract:

Green hydrogen is hydrogen produced using renewable energy sources. In the last few years, Oman aimed to reduce its dependency on fossil fuels. Recently, the hydrogen economy has become a global trend, and many countries have started to investigate the feasibility of implementing this sector. Oman created an alliance to establish the policy and rules for this sector. With motivation coming from both global and local interest in green hydrogen, this paper investigates the potential of producing hydrogen from wind and solar energies in three different locations in Oman, namely Duqm, Salalah, and Sohar. By using machine learning-based software “WEKA” and local metrological data, the project was designed to figure out which location has the highest wind and solar energy potential. First, various supervised models were tested to obtain their prediction accuracy, and it was found that the Random Forest (RF) model has the best prediction performance. The RF model was applied to 2021 metrological data for each location, and the results indicated that Duqm has the highest wind and solar energy potential. The system of one wind turbine in Duqm can produce 8335 MWh/year, which could be utilized in the water electrolysis process to produce 88847 kg of hydrogen mass, while a solar system consisting of 2820 solar cells is estimated to produce 1666.223 MWh/ year which is capable of producing 177591 kg of hydrogen mass.

Keywords: green hydrogen, machine learning, wind and solar energies, WEKA, supervised models, random forest

Procedia PDF Downloads 79

Authors: Vaishakh Medikeri

Abstract:

Lightning, the marvelous, spectacular and the awesome truth of nature is one of the greatest energy sources left unharnessed since ages. A single lightning bolt of lightning contains energy of about 15 billion joules. This huge amount of energy cannot be harnessed completely but partially. This paper proposes to harness the energy from lightning strikes. Throughout the globe the frequency of lightning is 40-50 flashes per second, totally 1.4 billion flashes per year; all of these flashes carrying an average energy of about 15 billion joules each. When a lightning bolt strikes the ground, tremendous amounts of energy is transferred to earth which propagates in the form of concentric circular energy waves. These waves have a frequency of about 7.83Hz. Harvesting the lightning bolt directly seems impossible, but harvesting the energy waves produced by the lightning is pretty easier. This can be done using a tricoil energy harnesser which is a new device which I have invented. We know that lightning bolt seeks the path which has minimum resistance down to the earth. For this we can make a lightning rod about 100 meters high. Now the lightning rod is attached to the tricoil energy harnesser. The tricoil energy harnesser contains three coils whose centers are collinear and all the coils are parallel to the ground. The first coil has one of its ends connected to the lightning rod and the other end grounded. There is a secondary coil wound on the first coil with one of its end grounded and the other end pointing to the ground and left unconnected and placed a little bit above the ground so that this end of the coil produces more intense currents, hence producing intense energy waves. The first coil produces very high magnetic fields and induces them in the second and third coils. Along with the magnetic fields induced by the first coil, the energy waves which are currents also flow through the second and the third coils. The second and the third coils are connected to a generator which in turn is connected to a capacitor which stores the electrical energy. The first coil is placed in the middle of the second and the third coil. The stored energy can be used for transmission of electricity. This new technique of harnessing the lightning strikes would be most efficient in places with more probability of the lightning strikes. Since we are using a lightning rod sufficiently long, the probability of cloud to ground strikes is increased. If the proposed apparatus is implemented, it would be a great source of pure and clean energy.

Keywords: generator, lightning rod, tricoil energy harnesser, harvesting energy

Procedia PDF Downloads 380

Authors: Jindong Gu, Matthias Schubert, Volker Tresp

Abstract:

In many outlier detection tasks, only training data belonging to one class, i.e., the positive class, is available. The task is then to predict a new data point as belonging either to the positive class or to the negative class, in which case the data point is considered an outlier. For this task, we propose a novel corrupted Generative Adversarial Network (CorGAN). In the adversarial process of training CorGAN, the Generator generates outlier samples for the negative class, and the Discriminator is trained to distinguish the positive training data from the generated negative data. The proposed framework is evaluated using an image dataset and a real-world network intrusion dataset. Our outlier-detection method achieves state-of-the-art performance on both tasks.

Keywords: one-class classification, outlier detection, generative adversary networks, semi-supervised learning

Procedia PDF Downloads 151

Authors: Zhiqing Zhang, Xiangong Zhou, Zihan Zhou

Abstract:

Ground fissures can be seen in some cities all over the world. As a special urban geological disaster, ground fissures in Xi'an have caused great harm to infrastructure. Chang'an Road Interchange in Xi'an City is a bridge across ground fissures. The damage to Chang'an Road interchange is the most serious and typical. To study the influence of ground fissures on the bridge, we established a bridge monitoring system. The main monitoring items include elevation monitoring, structural displacement monitoring, etc. The monitoring results show that the typical failure is mainly reflected in the bridge deck damage caused by horizontal tension and vertical dislocation. For the construction of urban interchange spanning ground fissures, the interchange should be divided reasonably, a simple support structure with less restriction should be adopted, and the monitoring of supports should be strengthened to prevent the occurrence of beam falling.

Keywords: bridge monitoring, ground fissures, typical disease, structural displacement

Procedia PDF Downloads 223

Authors: Xulu Yao, Moi Hoon Yap, Yanlong Zhang

Abstract:

As a branch of artificial neural network, deep learning is widely used in the field of image recognition, but the lack of its dataset leads to imperfect model learning. By analysing the data scale requirements of deep learning and aiming at the application in GUI generation, it is found that the collection of GUI dataset is a time-consuming and labor-consuming project, which is difficult to meet the needs of current deep learning network. To solve this problem, this paper proposes a semi-supervised deep learning model that relies on the original small-scale datasets to produce a large number of reliable data sets. By combining the cyclic neural network with the generated countermeasure network, the cyclic neural network can learn the sequence relationship and characteristics of data, make the generated countermeasure network generate reasonable data, and then expand the Rico dataset. Relying on the network structure, the characteristics of collected data can be well analysed, and a large number of reasonable data can be generated according to these characteristics. After data processing, a reliable dataset for model training can be formed, which alleviates the problem of dataset shortage in deep learning.

Keywords: GUI, deep learning, GAN, data augmentation

Procedia PDF Downloads 184

Authors: Ekeh Greg

Abstract:

Learning has been part of human existence from time immemorial. The aim of every learning is to know the truth. In education, it is desirable that true knowledge is imparted and imbibed. For this to be achieved, there is need for honesty, in this context, academic honesty among students, especially in e-learning. This is an ethical issue since honesty bothers on human conduct. However, research findings have shown that academic honesty has remained a big challenge to online learners, especially among the university students. This is worrisome since the university education is the final education system and a gateway to life in the wider society after schooling. If they are practicing honesty in their academic life, it is likely that they will practice honesty in the in the society, thereby bringing positive contributions to the society wherever they find themselves. With this in mind, the significance of this study becomes obvious. On grounds of this significance, this paper focuses on strategies that are adjudged certain to enhance the practice of honesty in e-learning so as to enable learners to be well equipped to contribute to the society through honest ways. The aim of the paper is to contribute to the efforts of instilling the consciousness and practice of honesty in the minds and hearts of learners. This will, in turn, promote effective teaching and learning, academic high standard, competence and self-confidence in university education. Philosophical methods of conceptual analysis, clarification, description and prescription are adopted for the study. Philosophical perspective is chosen so as to ground the paper on the basis of rationality rather than emotional sentiments and biases emanating from cultural, religious and ethnic differences and orientations. Such sentiments and biases can becloud objective reasoning and sound judgment. A review of related literature is also carried out. The findings show that academic honesty in e-learning is a cherished value, but it is bedeviled by some challenges, such as care-free attitude on the part of students and absence of monitoring. The findings also show that despite the challenges facing academic honesty, strategies such as self-discipline, determination, hard work, imbibing ethical and philosophical principles, among others, can certainly enhance the practice of honesty in e-learning among university students. The paper, therefore, concludes that these constitute strategies for enhancing academic honesty among students. Consequently, it is suggested that instructors, school counsellors and other stakeholders should endeavour to see that students are helped to imbibe these strategies and put them into practice. Students themselves are enjoined to cherish honesty in their academic pursuit and avoid short-cuts. Short-cuts can only lead to mediocrity and incompetence on the part of the learners, which may have long adverse consequences, both on themselves and others.

Keywords: academic, ethical, philosophical, strategies

Procedia PDF Downloads 76

Authors: Qin Long, Li Xiaoge

Abstract:

The Knowledge Graphs have now become a new form of knowledge representation. However, there is no consensus in regard to a plausible and definition of entities and relationships in the domain-specific knowledge graph. Further, in conjunction with several limitations and deficiencies, various domain-specific entities and relationships recognition approaches are far from perfect. Specifically, named entity recognition in Chinese domain is a critical task for the natural language process applications. However, a bottleneck problem with Chinese named entity recognition in new domains is the lack of annotated data. To address this challenge, a domain distant supervised named entity recognition framework is proposed. The framework is divided into two stages: first, the distant supervised corpus is generated based on the entity linking model of graph attention neural network; secondly, the generated corpus is trained as the input of the distant supervised named entity recognition model to train to obtain named entities. The link model is verified in the ccks2019 entity link corpus, and the F1 value is 2% higher than that of the benchmark method. The re-pre-trained BERT language model is added to the benchmark method, and the results show that it is more suitable for distant supervised named entity recognition tasks. Finally, it is applied in the computer field, and the results show that this framework can obtain domain named entities.

Keywords: distant named entity recognition, entity linking, knowledge graph, graph attention neural network

Procedia PDF Downloads 93

Authors: Sneha Thakur, Sanjeev Karmakar

Abstract:

This paper proposed hybrid Particle Swarm Optimization (PSO) – Long-Short Term Memory (LSTM) model for groundwater level prediction. The evaluation of the performance is realized using the parameters: root mean square error (RMSE) and mean absolute error (MAE). Ground water level forecasting will be very effective for planning water harvesting. Proper calculation of water level forecasting can overcome the problem of drought and flood to some extent. The objective of this work is to develop a ground water level forecasting model using deep learning technique integrated with optimization technique PSO by applying 29 years data of Chhattisgarh state, In-dia. It is important to find the precise forecasting in case of ground water level so that various water resource planning and water harvesting can be managed effectively.

Keywords: long short-term memory, particle swarm optimization, prediction, deep learning, groundwater level

Procedia PDF Downloads 78

Authors: Adam Reilly

Abstract:

The major claim of this paper is that both the English and Canadian branches of the common law have been ill-served by the 'institutional'/'remedial' taxonomy of constructive trusts; what shall be termed the 'orthodox taxonomy'.  The orthodox taxonomy is found both within the case law and the attendant academic commentary.  In truth, the orthodox taxonomy is especially dangerous because it contains a kernel of truth together with a misconception; the interplay of both has caused more harm than the misconception alone would have managed.  The kernel of truth is that some trusts arise automatically when the necessary facts occur ('institutional') and other trusts arise only by way of court order ('remedial').  The misconception is that these two labels represent an exhaustive nomenclature of two distinct 'kinds' of constructive trust such that any particular constructive trust must necessarily be 'institutional' if it is not 'remedial' and vice versa.  The central difficulty is that our understanding of 'remedial' trusts is relatively poor, with the result that anyone using the orthodox taxonomy shall be led astray in one of three ways: (i) by rejecting it wholesale; (ii) by adopting one ‘type’ of trust to the exclusion of the other (as in English law); or (iii) by applying it as an analytical device with sub-optimal results which are difficult to defend.  This paper shall seek to resolve these difficulties by clarifying the criteria for identifying and distinguishing true 'remedial' constructive trusts.  It shall then provide some working examples of how English and Canadian private law at present misunderstand constructive trusts and how that misunderstanding might be resolved once we distinguish the orthodox taxonomy's kernel of truth from the misconception outlined above.

Keywords: comparative law, constructive trusts, equitable remedies, remedial constructive trusts

Procedia PDF Downloads 141

Authors: Regina Chua

Abstract:

To reduce the costs of re-fielding a large survey, a Random Forest classifier was applied to measure the accuracy of classifying individuals into their assigned segments with the fewest possible questions. Given a long survey, one needed to determine the most predictive ten or fewer questions that would accurately assign new individuals to custom segments. Furthermore, the solution needed to be quick in its classification and usable in non-Python environments. In this paper, a supervised Random Forest classifier was modeled on a dataset with 7,000 individuals, 60 questions, and 254 features. The Random Forest consisted of an iterative collection of individual decision trees that result in a predicted segment with robust precision and recall scores compared to a single tree. A random 70-30 stratified sampling for training the algorithm was used, and accuracy trade-offs at different depths for each segment were identified. Ultimately, the Random Forest classifier performed at 87% accuracy at a depth of 10 with 20 instead of 254 features and 10 instead of 60 questions. With an acceptable accuracy in prioritizing feature selection, new tools were developed for non-Python environments: a worksheet with a formulaic version of the algorithm and an embedded function to predict the segment of an individual in real-time. Random Forest was determined to be an optimal classification model by its feature selection, performance, processing speed, and flexible application in other environments.

Keywords: machine learning, supervised learning, data science, random forest, classification, prediction, predictive modeling

Procedia PDF Downloads 94

Authors: Dharni Vasudhevan Venkatesan, K. E. Shanjay, H. Sujith Kumar, N. A. Abhilash, D. Aswin Ram, V. R. Sanal Kumar

Abstract:

Numerical studies on race car aerodynamics at wing in ground effect have been carried out using a steady 3d, double precision, pressure-based, and standard k-epsilon turbulence model. Through various parametric analytical studies we have observed that at a particular speed and ground clearance of the wings a favorable negative lift was found high at a particular angle of attack for all the physical models considered in this paper. The fact is that if the ground clearance height to chord length (h/c) is too small, the developing boundary layers from either side (the ground and the lower surface of the wing) can interact, leading to an altered variation of the aerodynamic characteristics at wing in ground effect. Therefore a suitable ground clearance must be predicted throughout the racing for a better performance of the race car, which obviously depends upon the coupled effects of the topography, wing orientation with respect to the ground, the incoming flow features and/or the race car speed. We have concluded that for the design of high performance and high speed race cars the adjustable wings capable to alter the ground clearance and the angles of attack is the best design option for any race car for racing safely with variable speeds.

Keywords: external aerodynamics, external flow choking, race car aerodynamics, wing in ground effect

Procedia PDF Downloads 356

Authors: S. S. Patil, Sachidanand Kini

Abstract:

Classification of land use patterns is compelling in complexity and variability of remote sensing imageries data. An imperative research in remote sensing application exploited to mine some of the significant spatially variable factors as land cover and land use from satellite images for remote arid areas in Karnataka State, India. The diverse classification techniques, unsupervised and supervised consisting of maximum likelihood, Mahalanobis distance, and minimum distance are applied in Bellary District in Karnataka State, India for the classification of the raw satellite images. The accuracy evaluations of results are compared visually with the standard maps with ground-truths. We initiated with the maximum likelihood technique that gave the finest results and both minimum distance and Mahalanobis distance methods over valued agriculture land areas. In meanness of mislaid few irrelevant features due to the low resolution of the satellite images, high-quality accord between parameters extracted automatically from the developed maps and field observations was found.

Keywords: Mahalanobis distance, minimum distance, supervised, unsupervised, user classification accuracy, producer's classification accuracy, maximum likelihood, kappa coefficient

Procedia PDF Downloads 183

Authors: Farid Khosravikia, Patricia Clayton

Abstract:

The main objective of this work is to evaluate the advantages and disadvantages of various machine learning techniques in two key steps of seismic hazard and risk assessment of different types of structures. The first step is the development of ground-motion models, which are used for forecasting ground-motion intensity measures (IM) given source characteristics, source-to-site distance, and local site condition for future events. IMs such as peak ground acceleration and velocity (PGA and PGV, respectively) as well as 5% damped elastic pseudospectral accelerations at different periods (PSA), are indicators of the strength of shaking at the ground surface. Typically, linear regression-based models, with pre-defined equations and coefficients, are used in ground motion prediction. However, due to the restrictions of the linear regression methods, such models may not capture more complex nonlinear behaviors that exist in the data. Thus, this study comparatively investigates potential benefits from employing other machine learning techniques as statistical method in ground motion prediction such as Artificial Neural Network, Random Forest, and Support Vector Machine. The results indicate the algorithms satisfy some physically sound characteristics such as magnitude scaling distance dependency without requiring pre-defined equations or coefficients. Moreover, it is shown that, when sufficient data is available, all the alternative algorithms tend to provide more accurate estimates compared to the conventional linear regression-based method, and particularly, Random Forest outperforms the other algorithms. However, the conventional method is a better tool when limited data is available. Second, it is investigated how machine learning techniques could be beneficial for developing probabilistic seismic demand models (PSDMs), which provide the relationship between the structural demand responses (e.g., component deformations, accelerations, internal forces, etc.) and the ground motion IMs. In the risk framework, such models are used to develop fragility curves estimating exceeding probability of damage for pre-defined limit states, and therefore, control the reliability of the predictions in the risk assessment. In this study, machine learning algorithms like artificial neural network, random forest, and support vector machine are adopted and trained on the demand parameters to derive PSDMs for them. It is observed that such models can provide more accurate estimates of prediction in relatively shorter about of time compared to conventional methods. Moreover, they can be used for sensitivity analysis of fragility curves with respect to many modeling parameters without necessarily requiring more intense numerical response-history analysis.

Keywords: artificial neural network, machine learning, random forest, seismic risk analysis, seismic hazard analysis, support vector machine

Procedia PDF Downloads 106

Authors: Soofiyan Atar, Adil Shaikh, Sahil Rajpurkar, Pragnesh Bhalala, Aniket Desai, Irfan Siddavatam

Abstract:

Deep reinforcement learning (deep RL) algorithms leverage the symbolic power of complex controllers by automating it by mapping sensory inputs to low-level actions. Deep RL eliminates the complex robot dynamics with minimal engineering. Deep RL provides high-risk involvement by directly implementing it in real-world scenarios and also high sensitivity towards hyperparameters. Tuning of hyperparameters on a pneumatic quadruped robot becomes very expensive through trial-and-error learning. This paper presents an automated learning control for a pneumatic quadruped robot using sample efficient deep Q learning, enabling minimal tuning and very few trials to learn the neural network. Long training hours may degrade the pneumatic cylinder due to jerk actions originated through stochastic weights. We applied this method to the pneumatic quadruped robot, which resulted in a hopping gait. In our process, we eliminated the use of a simulator and acquired a stable gait. This approach evolves so that the resultant gait matures more sturdy towards any stochastic changes in the environment. We further show that our algorithm performed very well as compared to programmed gait using robot dynamics.

Keywords: model-based reinforcement learning, gait stability, supervised learning, pneumatic quadruped

Procedia PDF Downloads 316

Authors: Yihao Kuang, Bowen Ding

Abstract:

With the increasing scale and complexity of knowledge graph, modern knowledge graph contains more and more types of entity, relationship, and attribute information. Therefore, in recent years, it has been a trend for knowledge graph inference to use reinforcement learning to deal with large-scale, incomplete, and noisy knowledge graph and improve the inference effect and interpretability. The Proximal Policy Optimization (PPO) algorithm utilizes a near-end strategy optimization approach. This allows for more extensive updates of policy parameters while constraining the update extent to maintain training stability. This characteristic enables PPOs to converge to improve strategies more rapidly, often demonstrating enhanced performance early in the training process. Furthermore, PPO has the advantage of offline learning, effectively utilizing historical experience data for training and enhancing sample utilization. This means that even with limited resources, PPOs can efficiently train for reinforcement learning tasks. Based on these characteristics, this paper aims to obtain better and more efficient inference effect by introducing PPO into knowledge inference technology.

Keywords: reinforcement learning, PPO, knowledge inference, supervised learning

Procedia PDF Downloads 67

Authors: Vishnu Pratap Singh Kirar

Abstract:

Artificial Neural Network (ANN) can be trained using backpropagation (BP). It is the most widely used algorithm for supervised learning with multi-layered feed-forward networks. Efficient learning by the BP algorithm is required for many practical applications. The BP algorithm calculates the weight changes of artificial neural networks, and a common approach is to use a two-term algorithm consisting of a learning rate (LR) and a momentum factor (MF). The major drawbacks of the two-term BP learning algorithm are the problems of local minima and slow convergence speeds, which limit the scope for real-time applications. Recently the addition of an extra term, called a proportional factor (PF), to the two-term BP algorithm was proposed. The third increases the speed of the BP algorithm. However, the PF term also reduces the convergence of the BP algorithm, and criteria for evaluating convergence are required to facilitate the application of the three terms BP algorithm. Although these two seem to be closely related, as described later, we summarize various improvements to overcome the drawbacks. Here we compare the different methods of convergence of the new three-term BP algorithm.

Keywords: neural network, backpropagation, local minima, fast convergence rate

Procedia PDF Downloads 498

Authors: Paulo Cesar Pereira Junior, Alexandre Monteiro, Rafael da Luz Ribeiro, Antonio Carlos Sobieranski, Aldo von Wangenheim

Abstract:

In this paper, we present a comparison between convolutional neural networks and classical computer vision approaches, for the specific precision agriculture problem of weed mapping on sugarcane fields aerial images. A systematic literature review was conducted to find which computer vision methods are being used on this specific problem. The most cited methods were implemented, as well as four models of convolutional neural networks. All implemented approaches were tested using the same dataset, and their results were quantitatively and qualitatively analyzed. The obtained results were compared to a human expert made ground truth for validation. The results indicate that the convolutional neural networks present better precision and generalize better than the classical models.

Keywords: convolutional neural networks, deep learning, digital image processing, precision agriculture, semantic segmentation, unmanned aerial vehicles

Procedia PDF Downloads 260

Authors: Gaokai Liu

Abstract:

Automatic defect detection for nanomaterial images is widely required in industrial scenarios. Deep learning approaches are considered as the most effective solutions for the great majority of image-based tasks. In this paper, an edge guidance network for defect segmentation is proposed. First, the encoder path with multiple convolution and downsampling operations is applied to the acquisition of shared features. Then two decoder paths both are connected to the last convolution layer of the encoder and supervised by the edge and segmentation labels, respectively, to guide the whole training process. Meanwhile, the edge and encoder outputs from the same stage are concatenated to the segmentation corresponding part to further tune the segmentation result. Finally, the effectiveness of the proposed method is verified via the experiments on open nanofibrous datasets.

Keywords: deep learning, defect detection, image segmentation, nanomaterials

Procedia PDF Downloads 149

Authors: Brindha Senthil Kumar, Payel Chakraborty, Senthil Kumar Nachimuthu, Arindam Maitra, Prem Nath

Abstract:

Gastric cancer is predominantly caused by demographic and diet factors as compared to other cancer types. The aim of the study is to predict Early Gastric Cancer (ECG) from diet and lifestyle factors using supervised machine learning algorithms. For this study, 160 healthy individual and 80 cases were selected who had been followed for 3 years (2016-2019), at Civil Hospital, Aizawl, Mizoram. A dataset containing 11 features that are core risk factors for the gastric cancer were extracted. Supervised machine algorithms: Logistic Regression, Naive Bayes, Support Vector Machine (SVM), Multilayer perceptron, and Random Forest were used to analyze the dataset using Python Jupyter Notebook Version 3. The obtained classified results had been evaluated using metrics parameters: minimum_false_positives, brier_score, accuracy, precision, recall, F1_score, and Receiver Operating Characteristics (ROC) curve. Data analysis results showed Naive Bayes - 88, 0.11; Random Forest - 83, 0.16; SVM - 77, 0.22; Logistic Regression - 75, 0.25 and Multilayer perceptron - 72, 0.27 with respect to accuracy and brier_score in percent. Naive Bayes algorithm out performs with very low false positive rates as well as brier_score and good accuracy. Naive Bayes algorithm classification results in predicting ECG showed very satisfactory results using only diet cum lifestyle factors which will be very helpful for the physicians to educate the patients and public, thereby mortality of gastric cancer can be reduced/avoided with this knowledge mining work.

Keywords: Early Gastric cancer, Machine Learning, Diet, Lifestyle Characteristics

Procedia PDF Downloads 161

Authors: Tauhidur Rahman, Gitartha Kalita

Abstract:

In this study a four span box girder bridge is considered and effect of the rotational and translational earthquake ground motion have been thoroughly investigated. This study is motivated by the fact that in many countries the translational and rotational components of earthquake ground motion, especially rocking, is not adequately considered in analysing the overall response of the structures subjected to earthquake ground excitations. Much consideration is given to only the horizontal components of the earthquake ground motion during the response analysis of structures. In the present research work, P waves, SV waves and Rayleigh wave excitations are considered for different angle of incidence. In the present paper, the four span bridge is model considering the effects of vertical and rocking components of P, SV and Rayleigh wave excitations. Ground responses namely displacement, velocity and acceleration of the substructures of the bridge have been considered for rotational and translational effects in addition to the horizontal ground motion due to earthquake and wind.

Keywords: ground motion, response, rotational effects, translational effects

Procedia PDF Downloads 447

Authors: Akintunde Alo

Abstract:

The incessant encroachments into the forest ecosystem by the farmers and local contractors constitute a major threat to the conservation of genetic resources and biodiversity in Nigeria. To propose a viable monitoring system, this study employed Geographic Information System (GIS) technology to assess the changes that occurred for a period of five years (between 2011 and 2016) in Ikere forest reserve. Landsat imagery of the forest reserve was obtained. For the purpose of geo-referencing the acquired satellite imagery, ground-truth coordinates of some benchmark places within the forest reserve was relied on. Supervised classification algorithm, image processing, vectorization and map production were realized using ArcGIS. Various land use systems within the forest ecosystem were digitized into polygons of different types and colours for 2011 and 2016, roads were represented with lines of different thickness and colours. Of the six land-use delineated, the grassland increased from 26.50 % in 2011 to 45.53% in 2016 of the total land area with a percentage change of 71.81 %. Plantations of Gmelina arborea and Tectona grandis on the other hand reduced from 62.16 % in 2011 to 27.41% in 2016. The farmland and degraded land recorded percentage change of about 176.80 % and 8.70 % respectively from 2011 to 2016. Overall, the rate of deforestation in the study area is on the increase and becoming severe. About 72.59% of the total land area has been converted to non-forestry uses while the remnant 27.41% is occupied by plantations of Gmelina arborea and Tectona grandis. Interestingly, over 55 % of the plantation area in 2011 has changed to grassland, or converted to farmland and degraded land in 2016. The rate of change over time was about 9.79 % annually. Based on the results, rapid actions to prevail on the encroachers to stop deforestation and encouraged re-afforestation in the study area are recommended.

Keywords: land use change, forest reserve, satellite imagery, geographical information system

Procedia PDF Downloads 356