Search results for: deep submicronCMOS circuits

Authors: Abobaker Mohammed Qasem Farhan, Shangming Yang, Mohammed Al-Nehari

Abstract:

Pneumonia is a kind of lung disease that creates congestion in the chest. Such pneumonic conditions lead to loss of life due to the severity of high congestion. Pneumonic lung disease is caused by viral pneumonia, bacterial pneumonia, or COVID-19 induced pneumonia. The early prediction and classification of such lung diseases help reduce the mortality rate. We propose the automatic Computer-Aided Diagnosis (CAD) system in this paper using the deep learning approach. The proposed CAD system takes input from raw computerized tomography (CT) scans of the patient's chest and automatically predicts disease classification. We designed the Hybrid Deep Learning Algorithm (HDLA) to improve accuracy and reduce processing requirements. The raw CT scans are pre-processed first to enhance their quality for further analysis. We then applied a hybrid model that consists of automatic feature extraction and classification. We propose the robust 2D Convolutional Neural Network (CNN) model to extract the automatic features from the pre-processed CT image. This CNN model assures feature learning with extremely effective 1D feature extraction for each input CT image. The outcome of the 2D CNN model is then normalized using the Min-Max technique. The second step of the proposed hybrid model is related to training and classification using different classifiers. The simulation outcomes using the publicly available dataset prove the robustness and efficiency of the proposed model compared to state-of-art algorithms.

Keywords: CT scans, COVID-19, deep learning, image processing, pneumonia, lung disease.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 518

Authors: Tonislav Ivanov, Oleksii Nedashkivskyi, Denis Babeshko, Vadim Pinskiy, Matthew Putman

Abstract:

Training a machine learning model for object detection that generalizes well is known to benefit from a training dataset with diverse examples. However, training datasets usually contain many repeats of common examples of a class and lack rarely seen examples. This is due to the process commonly used during human annotation where a person would proceed sequentially through a list of images labeling a sufficiently high total number of examples. Instead, the method presented involves an active process where, after the initial labeling of several images is completed, the next subset of images for labeling is selected by an algorithm. This process of algorithmic image selection and manual labeling continues in an iterative fashion. The algorithm used for the image selection is a deep learning algorithm, based on the U-shaped architecture, which quantifies the presence of unseen data in each image in order to find images that contain the most novel examples. Moreover, the location of the unseen data in each image is highlighted, aiding the labeler in spotting these examples. Experiments performed using semiconductor wafer data show that labeling a subset of the data, curated by this algorithm, resulted in a model with a better performance than a model produced from sequentially labeling the same amount of data. Also, similar performance is achieved compared to a model trained on exhaustive labeling of the whole dataset. Overall, the proposed approach results in a dataset that has a diverse set of examples per class as well as more balanced classes, which proves beneficial when training a deep learning model.

Keywords: Computer vision, deep learning, object detection, semiconductor.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 774

Authors: Yifan Fan, Xudong Luo, Pingping Lin

Abstract:

An essential task in the field of artificial intelligence is to allow computers to interact with people through natural language. Therefore, researches such as virtual assistants and dialogue systems have received widespread attention from industry and academia. The response generation plays a crucial role in dialogue systems, so to push forward the research on this topic, this paper surveys various methods for response generation. We sort out these methods into three categories. First one includes finite state machine methods, framework methods, and instance methods. The second contains full-text indexing methods, ontology methods, vast knowledge base method, and some other methods. The third covers retrieval methods and generative methods. We also discuss some hybrid methods based knowledge and deep learning. We compare their disadvantages and advantages and point out in which ways these studies can be improved further. Our discussion covers some studies published in leading conferences such as IJCAI and AAAI in recent years.

Keywords: Retrieval, generative, deep learning, response generation, knowledge.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1163

Authors: Shakil A. Kagzi, Sanjay Patil, Harit K. Raval

Abstract:

Tailor Welded Blanks (TWB) are utilized in automotive industries widely because of their advantage of weight and cost reduction and maintaining required strength and structural integrity. TWB consist of two or more sheet having dissimilar or similar material and thickness; welded together to form a single sheet before forming it to desired shape. Forming of the tailor welded blank is affected by ratio of thickness of blanks, ratio of their strength, etc. mainly due to in-homogeneity of material. In the present work the relative effect of these parameters on weld line movement is studied during deep drawing of TWB using FE simulation using HYPERWORKS. The simulation is validated with results from the literature. Simulations were than performed based on Taguchi orthogonal array followed by the ANOVA analysis to determine the significance of these parameters on forming of TWB.

Keywords: ANOVA, Deep drawing, Tailor Welded Blank, TWB, Weld line movement.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2765

Authors: Behzad Khademian, Mohammad Haeri

Abstract:

In this paper, an effective sliding mode design is applied to chaos synchronization. The proposed controller can make the states of two identical modified Chua-s circuits globally asymptotically synchronized. Numerical results are provided to show the effectiveness and robustness of the proposed method.

Keywords: Sliding mode, Chaos synchronization, Modified Chua's circuit.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1264

Authors: Ashwani K. Rana, Narottam Chand, Vinod Kapoor

Abstract:

Aggressive scaling of MOS devices requires use of ultra-thin gate oxides to maintain a reasonable short channel effect and to take the advantage of higher density, high speed, lower cost etc. Such thin oxides give rise to high electric fields, resulting in considerable gate tunneling current through gate oxide in nano regime. Consequently, accurate analysis of gate tunneling current is very important especially in context of low power application. In this paper, a simple and efficient analytical model has been developed for channel and source/drain overlap region gate tunneling current through ultra thin gate oxide n-channel MOSFET with inevitable deep submicron effect (DSME).The results obtained have been verified with simulated and reported experimental results for the purpose of validation. It is shown that the calculated tunnel current is well fitted to the measured one over the entire oxide thickness range. The proposed model is suitable enough to be used in circuit simulator due to its simplicity. It is observed that neglecting deep sub-micron effect may lead to large error in the calculated gate tunneling current. It is found that temperature has almost negligible effect on gate tunneling current. It is also reported that gate tunneling current reduces with the increase of gate oxide thickness. The impact of source/drain overlap length is also assessed on gate tunneling current.

Keywords: Gate tunneling current, analytical model, gate dielectrics, non uniform poly gate doping, MOSFET, fringing field effect and image charges.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1701

Authors: Jimeng Shi, Mahek Jain, Giri Narasimhan

Abstract:

Time Series Forecasting (TSF) is used to predict the target variables at a future time point based on the learning from previous time points. To keep the problem tractable, learning methods use data from a fixed length window in the past as an explicit input. In this paper, we study how the performance of predictive models change as a function of different look-back window sizes and different amounts of time to predict into the future. We also consider the performance of the recent attention-based transformer models, which had good success in the image processing and natural language processing domains. In all, we compare four different deep learning methods (Recurrent Neural Network (RNN), Long Short-term Memory (LSTM), Gated Recurrent Units (GRU), and Transformer) along with a baseline method. The dataset (hourly) we used is the Beijing Air Quality Dataset from the website of University of California, Irvine (UCI), which includes a multivariate time series of many factors measured on an hourly basis for a period of 5 years (2010-14). For each model, we also report on the relationship between the performance and the look-back window sizes and the number of predicted time points into the future. Our experiments suggest that Transformer models have the best performance with the lowest Mean   Absolute Errors (MAE = 14.599, 23.273) and Root Mean Square Errors (RSME = 23.573, 38.131) for most of our single-step and multi-steps predictions. The best size for the look-back window to predict 1 hour into the future appears to be one day, while 2 or 4 days perform the best to predict 3 hours into the future.

Keywords: Air quality prediction, deep learning algorithms, time series forecasting, look-back window.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1071

Authors: C. Virgilio, J. Carlier, P. Campistron, M. Toubal, P. Garnier, L. Broussous, V. Thomy, B. Nongaillard

Abstract:

Wetting efficiency of microstructures or nanostructures patterned on Si wafers is a real challenge in integrated circuits manufacturing. In fact, bad or non-uniform wetting during wet processes limits chemical reactions and can lead to non-complete etching or cleaning inside the patterns and device defectivity. This issue is more and more important with the transistors size shrinkage and concerns mainly high aspect ratio structures. Deep Trench Isolation (DTI) structures enabling pixels’ isolation in imaging devices are subject to this phenomenon. While low-frequency acoustic reflectometry principle is a well-known method for Non Destructive Test applications, we have recently shown that it is also well suited for nanostructures wetting characterization in a higher frequency range. In this paper, we present a high-frequency acoustic reflectometry characterization of DTI wetting through a confrontation of both experimental and modeling results. The acoustic method proposed is based on the evaluation of the reflection of a longitudinal acoustic wave generated by a 100 µm diameter ZnO piezoelectric transducer sputtered on the silicon wafer backside using MEMS technologies. The transducers have been fabricated to work at 5 GHz corresponding to a wavelength of 1.7 µm in silicon. The DTI studied structures, manufactured on the wafer frontside, are crossing trenches of 200 nm wide and 4 µm deep (aspect ratio of 20) etched into a Si wafer frontside. In that case, the acoustic signal reflection occurs at the bottom and at the top of the DTI enabling its characterization by monitoring the electrical reflection coefficient of the transducer. A Finite Difference Time Domain (FDTD) model has been developed to predict the behavior of the emitted wave. The model shows that the separation of the reflected echoes (top and bottom of the DTI) from different acoustic modes is possible at 5 Ghz. A good correspondence between experimental and theoretical signals is observed. The model enables the identification of the different acoustic modes. The evaluation of DTI wetting is then performed by focusing on the first reflected echo obtained through the reflection at Si bottom interface, where wetting efficiency is crucial. The reflection coefficient is measured with different water / ethanol mixtures (tunable surface tension) deposited on the wafer frontside. Two cases are studied: with and without PFTS hydrophobic treatment. In the untreated surface case, acoustic reflection coefficient values with water show that liquid imbibition is partial. In the treated surface case, the acoustic reflection is total with water (no liquid in DTI). The impalement of the liquid occurs for a specific surface tension but it is still partial for pure ethanol. DTI bottom shape and local pattern collapse of the trenches can explain these incomplete wetting phenomena. This high-frequency acoustic method sensitivity coupled with a FDTD propagative model thus enables the local determination of the wetting state of a liquid on real structures. Partial wetting states for non-hydrophobic surfaces or low surface tension liquids are then detectable with this method.

Keywords: Wetting, acoustic reflectometry, gigahertz, semiconductor.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1273

Authors: Xiaocong Liu, Huazhen Wang, Ting He, Xiaozheng Li, Weihan Zhang, Jian Chen

Abstract:

The utilization of electronic medical record (EMR) data to establish the disease diagnosis model has become an important research content of biomedical informatics. Deep learning can automatically extract features from the massive data, which brings about breakthroughs in the study of EMR data. The challenge is that deep learning lacks semantic knowledge, which leads to impracticability in medical science. This research proposes a method of incorporating lexical-semantic knowledge from abundant entities into a convolutional neural network (CNN) framework for pediatric disease diagnosis. Firstly, medical terms are vectorized into Lexical Semantic Vectors (LSV), which are concatenated with the embedded word vectors of word2vec to enrich the feature representation. Secondly, the semantic distribution of medical terms serves as Semantic Decision Guide (SDG) for the optimization of deep learning models. The study evaluates the performance of LSV-SDG-CNN model on four kinds of Chinese EMR datasets. Additionally, CNN, LSV-CNN, and SDG-CNN are designed as baseline models for comparison. The experimental results show that LSV-SDG-CNN model outperforms baseline models on four kinds of Chinese EMR datasets. The best configuration of the model yielded an F1 score of 86.20%. The results clearly demonstrate that CNN has been effectively guided and optimized by lexical-semantic knowledge, and LSV-SDG-CNN model improves the disease classification accuracy with a clear margin.

Keywords: lexical semantics, feature representation, semantic decision, convolutional neural network, electronic medical record

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 547

Authors: Harpreet Singh

Abstract:

Finding relevant information on the World Wide Web is becoming highly challenging day by day. Web usage mining is used for the extraction of relevant and useful knowledge, such as user behaviour patterns, from web access log records. Web access log records all the requests for individual files that the users have requested from the website. Web usage mining is important for Customer Relationship Management (CRM), as it can ensure customer satisfaction as far as the interaction between the customer and the organization is concerned. Web usage mining is helpful in improving website structure or design as per the user’s requirement by analyzing the access log file of a website through a log analyzer tool. The focus of this paper is to enhance the accessibility and usability of a guitar selling web site by analyzing their access log through Deep Log Analyzer tool. The results show that the maximum number of users is from the United States and that they use Opera 9.8 web browser and the Windows XP operating system.

Keywords: Web usage mining, log file, web mining, data mining, deep log analyser.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1032

Authors: Chun-Wei Huang, Jyh-Cherng Gu, Ming-Ta Yang

Abstract:

Non-synchronous breakage or line failure in power systems with light or no loads can lead to core saturation in transformers or potential transformers. This can cause component and capacitance matching resulting in the formation of resonant circuits, which trigger ferroresonance. This study employed a wavelet transform for the detection of ferroresonance. Simulation results demonstrate the efficacy of the proposed method.

Keywords: Ferroresonance, Wavelet Transform, Intelligent Electronic Device, Transformer.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1628

Authors: Rômulo Volpato, Filipe Ramos, Paulo Crepaldi, Michel Santana, Tales C Pimenta

Abstract:

This work presents an approach for the measurement of mutual inductance on near field inductive coupling. The mutual inductance between inductive circuits allows the simulation of energy transfer from reader to tag, that can be used in RFID and powerless implantable devices. It also allows one to predict the maximum voltage in the tag of the radio-frequency system.

Keywords: RFID, Inductive Coupling, Energy Transfer, Implantable Device

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2294

Authors: Samiah Alammari, Nassim Ammour

Abstract:

When providing a massive number of tasks successively to a deep learning process, a good performance of the model requires preserving the previous tasks data to retrain the model for each upcoming classification. Otherwise, the model performs poorly due to the catastrophic forgetting phenomenon. To overcome this shortcoming, we developed a successful continual learning deep model for remote sensing hyperspectral image regions classification. The proposed neural network architecture encapsulates two trainable subnetworks. The first module adapts its weights by minimizing the discrimination error between the land-cover classes during the new task learning, and the second module tries to learn how to replicate the data of the previous tasks by discovering the latent data structure of the new task dataset. We conduct experiments on hyperspectral image (HSI) dataset on Indian Pines. The results confirm the capability of the proposed method.

Keywords: Continual learning, data reconstruction, remote sensing, hyperspectral image segmentation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 178

Authors: A. Rabin, G. Mazor, I. Ladizhenski, R. Z. Shneck

Abstract:

The hardness and hardenability of AISI D2 cold work tool steel with conventional quenching (CQ), deep cryogenic quenching (DCQ) and rapid deep cryogenic quenching heat treatments caused by temporary porous coating based on magnesium sulfate was investigated. Each of the cooling processes was examined from the perspective of the full process efficiency, heat flux in the austenite-martensite transformation range followed by characterization of the temporary porous layer made of magnesium sulfate using confocal laser scanning microscopy (CLSM), surface and core hardness and hardenability using Vickers hardness technique. The results show that the cooling rate (CR) at the austenite-martensite transformation range has a high influence on the hardness of the studied steel.

Keywords: AISI D2, controllable cooling, magnesium sulfate coating, rapid cryogenic heat treatment, temporary porous layer.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 317

Authors: Bharatendra Rai

Abstract:

Sequences of words in text data have long-term dependencies and are known to suffer from vanishing gradient problem when developing deep learning models. Although recurrent networks such as long short-term memory networks help overcome this problem, achieving high text classification performance is a challenging problem. Convolutional recurrent networks that combine advantages of long short-term memory networks and convolutional neural networks, can be useful for text classification performance improvements. However, arriving at suitable hyperparameter values for convolutional recurrent networks is still a challenging task where fitting of a model requires significant computing resources. This paper illustrates the advantages of using convolutional recurrent networks for text classification with the help of statistically planned computer experiments for hyperparameter tuning. 

Keywords: Convolutional recurrent networks, hyperparameter tuning, long short-term memory networks, Tukey honest significant differences

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 61

Authors: Eslamizadeh, Azat., Samanirad, Shahram

Abstract:

The Yazd-Ardakan basin in Central Iran has two separated aquifers. The shallow unconfined aquifer is supplies 40 Qanats. The deep saturated confined aquifer is the main water storage. Due to over-withdrawal, water table has been decreasing during last 25 years. Recent study shows that the shortage of the aquifer is about 16 meters and land subsidence is 0.5 - 1.2 meters. Long deep cracks are found just above the aquifer and devour the irrigation water and floods. Although the most cracks direction is NW-SE and could be compared to the main direction of YA basin, there is no direct evidence for relation between land subsidence and the huge cracks. Large-scale water pumping has been decreased the water pressure in aquifer. The pressure decline disturbed the balance and increased the pressure of overlying sediments. So porosity decreased and compaction started. Then, sediments compaction developed and made land subsidence and some huge cracks slowly.

Keywords: Land subsidence, Iran, Yazd, aquifer

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1781

Authors: Thinh Ngo, Paul Rad, Brian Kelley

Abstract:

Noise estimation is essential in today wireless systems for power control, adaptive modulation, interference suppression and quality of service. Deep learning (DL) has already been applied in the physical layer for modulation and signal classifications. Unacceptably low accuracy of less than 50% is found to undermine traditional application of DL classification for SNR prediction. In this paper, we use divide-and-conquer algorithm and classifier fusion method to simplify SNR classification and therefore enhances DL learning and prediction. Specifically, multiple CNNs are used for classification rather than a single CNN. Each CNN performs a binary classification of a single SNR with two labels: less than, greater than or equal. Together, multiple CNNs are combined to effectively classify over a range of SNR values from −20 ≤ SNR ≤ 32 dB.We use pre-trained CNNs to predict SNR over a wide range of joint channel parameters including multiple Doppler shifts (0, 60, 120 Hz), power-delay profiles, and signal-modulation types (QPSK,16QAM,64-QAM). The approach achieves individual SNR prediction accuracy of 92%, composite accuracy of 70% and prediction convergence one order of magnitude faster than that of traditional estimation.

Keywords: Classification, classifier fusion, CNN, Deep Learning, prediction, SNR.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 670

Authors: Subodh Thankachan, Manisha Pattanaik, S. S. Rajput

Abstract:

In this paper, a new BiCMOS CCII and CCCII, capable of operate at ±0.5V and having wide dynamic range with achieved bandwidth of 480MHz and 430MHz respectively have been proposed. The structures have been found to be insensitive to the threshold voltage variations. The proposed circuits are suitable for implementation using 0.25μm BiCMOS technology. Pspice simulations confirm the performance of the proposed structures.

Keywords: BiCMOS, Current conveyor, Compound current conveyor, Low supply voltage, Threshold voltage variation.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1654

Authors: Ming-Jen Huang, Chun-Fang Huang, Chiching Wei

Abstract:

With the recent advance of the deep neural network, we observe new applications of NLP (natural language processing) and CV (computer vision) powered by deep neural networks for processing business documents. However, creating a real-world document processing system needs to integrate several NLP and CV tasks, rather than treating them separately. There is a need to have a unified approach for processing documents containing textual and graphical elements with rich formats, diverse layout arrangement, and distinct semantics. In this paper, a framework that fulfills this unified approach is presented. The framework includes a representation model definition for holding the information generated by various tasks and specifications defining the coordination between these tasks. The framework is a blueprint for building a system that can process documents with rich formats, styles, and multiple types of elements. The flexible and lightweight design of the framework can help build a system for diverse business scenarios, such as contract monitoring and reviewing.

Keywords: Document processing, framework, formal definition, machine learning.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 599

Authors: E. Oke, A. O. Ladokun, J. O. Daramola, O. M. Onagbesan

Abstract:

There has been a growing interest on the effects of access to pasture on poultry health status. However, there is a paucity of data on the relative benefits of grass and legume pastures. An experiment was conducted to determine the effects of rearing systems {deep litter system (DL), deep litter with access to legumes (LP) or grass (GP) pastures} haematology and serum chemistry of ISA Brown layers. The study involved the use of two hundred and forty 12 weeks old pullets. The birds were reared until 60 weeks of age. Eighty birds were assigned to each treatment; each treatment had four replicates of 20 birds each. Blood samples (2.5 ml) were collected from the wing vein of two birds per replicate and serum chemistry and haematological parameters were determined. The results showed that there were no significant differences between treatments in all the parameters considered at 18 weeks of age. At 24 weeks old, the percentage of heterophyl (HET) in DL and LP were similar but higher than that of GP. The ratio of H:L was higher (P<0.05) in DL than those of LP and GP while LP and GP were comparable. At week 38 of age, the percentage of PCV in the birds in LP and GP were similar but the birds in DL had significantly lower level than that of GP. In the early production phase, serum total protein of the birds in LP was similar to that of GP but higher (P<0.05) than that of DL. At the peak production phase (week 38), the total protein in GP and DL were similar but significantly lower than that of LP. The albumin level in LP was greater (P<0.05) than GP but similar to that of DL. In the late production phase, the total protein in LP was significantly higher than that of DL but similar to that of GP. It was concluded that rearing chickens in either grass or legume pasture did not have deleterious effects on the health of laying chickens but improved some parameters including blood protein and HET/lymphocyte.

Keywords: Rearing systems, Stylosanthes, Cynodon serum chemistry, haematology, hen.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2030

Authors: Jia Xin Low, Keng Wah Choo

Abstract:

This paper presents an automatic normal and abnormal heart sound classification model developed based on deep learning algorithm. MITHSDB heart sounds datasets obtained from the 2016 PhysioNet/Computing in Cardiology Challenge database were used in this research with the assumption that the electrocardiograms (ECG) were recorded simultaneously with the heart sounds (phonocardiogram, PCG). The PCG time series are segmented per heart beat, and each sub-segment is converted to form a square intensity matrix, and classified using convolutional neural network (CNN) models. This approach removes the need to provide classification features for the supervised machine learning algorithm. Instead, the features are determined automatically through training, from the time series provided. The result proves that the prediction model is able to provide reasonable and comparable classification accuracy despite simple implementation. This approach can be used for real-time classification of heart sounds in Internet of Medical Things (IoMT), e.g. remote monitoring applications of PCG signal.

Keywords: Convolutional neural network, discrete wavelet transform, deep learning, heart sound classification.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1121

Authors: G. Türkmen, Ö. Tekin, K. Kurtuluş, Y. Y. Yurtseven, M. Baran

Abstract:

A scanning electron microscope (SEM) is a type of electron microscope mainly used in nanoscience and nanotechnology areas. Automatic image recognition and classification are among the general areas of application concerning SEM. In line with these usages, the present paper proposes a deep learning algorithm that classifies SEM images into nine categories by means of an online application to simplify the process. The NFFA-EUROPE - 100% SEM data set, containing approximately 21,000 images, was used to train and test the algorithm at 80% and 20%, respectively. Validation was carried out using a separate data set obtained from the Middle East Technical University (METU) in Turkey. To increase the accuracy in the results, the Inception ResNet-V2 model was used in view of the Fine-Tuning approach. By using a confusion matrix, it was observed that the coated-surface category has a negative effect on the accuracy of the results since it contains other categories in the data set, thereby confusing the model when detecting category-specific patterns. For this reason, the coated-surface category was removed from the train data set, hence increasing accuracy by up to 96.5%.

Keywords: Convolutional Neural Networks, deep learning, image classification, scanning electron microscope.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 149

Authors: Hemani Kaushal, V.K.Jain, Subrat Kar

Abstract:

The performance of ground to deep space optical communication systems is degraded by distortion of the beam as it propagates through the turbulent atmosphere. Turbulence causes fluctuations in the intensity of the received signal which ultimately affects the acquisition time required to acquire and locate the spaceborne target using narrow laser beam. In this paper, performance of free-space optical (FSO) communication system in atmospheric turbulence has been analyzed in terms of acquisition time for coherent and non-coherent modulation schemes. Numerical results presented in graphical and tabular forms show that the acquisition time increases with the increase in turbulence level. This is true for both schemes. The BPSK has lowest acquisition time among all schemes. In non-coherent schemes, M-PPM performs better than the other schemes. With the increase in M, acquisition time becomes lower, but at the cost of increase in system complexity.

Keywords: Atmospheric Turbulence, Acquisition Time, BinaryPhase Shift Keying (BPSK), Free-Space Optical (FSO)Communication System, M-ary Pulse Position Modulation (M-PPM), Coherent/Non-coherent Modulation Schemes.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1746

Authors: Zhifeng Kong

Abstract:

Over-parameterized neural networks have attracted a great deal of attention in recent deep learning theory research, as they challenge the classic perspective of over-fitting when the model has excessive parameters and have gained empirical success in various settings. While a number of theoretical works have been presented to demystify properties of such models, the convergence properties of such models are still far from being thoroughly understood. In this work, we study the convergence properties of training two-hidden-layer partially over-parameterized fully connected networks with the Rectified Linear Unit activation via gradient descent. To our knowledge, this is the first theoretical work to understand convergence properties of deep over-parameterized networks without the equally-wide-hidden-layer assumption and other unrealistic assumptions. We provide a probabilistic lower bound of the widths of hidden layers and proved linear convergence rate of gradient descent. We also conducted experiments on synthetic and real-world datasets to validate our theory.

Keywords: Over-parameterization, Rectified Linear Units (ReLU), convergence, gradient descent, neural networks.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 843

Authors: E. De Finis, P. Gattinoni, L. Scesi

Abstract:

Massive rock avalanches formed some of the largest landslide deposits on Earth and they represent one of the major geohazards in high-relief mountains. This paper interprets a very large sedimentary fan (the Sernio fan, Valtellina, Northern Italy), located 20 Km SW from Val Pola Rock avalanche (1987), as the deposit of a partial collapse of a Deep Seated Gravitational Slope Deformation (DSGSD), afterwards eroded and buried by debris flows. The proposed emplacement sequence has been reconstructed based on geomorphological, structural and mechanical evidences. The Sernio fan is actually considered anomalous with reference to the very high ratio between the fan area (≈ 4.5km2) and the basin area (≈ 3km2). The morphology of the fan area is characterised by steep slopes (dip ≈ 20%) and the fan apex is extended for 1.8 km inside the small catchment basin. This sedimentary fan was originated by a landslide that interested a part of a large deep-seated gravitational slope deformation, involving a wide area of about 55 km². The main controlling factor is tectonic and it is related to the proximity to regional fault systems and the consequent occurrence of fault weak rocks (GSI locally lower than 10 with compressive stress lower than 20MPa). Moreover, the fan deposit shows sedimentary evidences of recent debris flow events. The best current explanation of the Sernio fan involves an initial failure of some hundreds of Mm3. The run-out was quite limited because of the morphology of Valtellina’s valley floor, and the deposit filled the main valley forming a landslide dam, as confirmed by the lacustrine deposits detected upstream the fan. Nowadays the debris flow events represent the main hazard in the study area.

Keywords: Anomalous sedimentary fans, debris flow, deep seated gravitational slope deformation, Italy, rock avalanche.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1714

Authors: Hanif Che Lah, Wee Leong Son, Rozita Borhan

Abstract:

This paper is about method to produce a stable and accurate constant output pulse width regardless of the amplitude, period and pulse width variation of the input signal source. The pulse generated is usually being used in numerous applications as the reference input source to other circuits in the system. Therefore, it is crucial to produce a clean and constant pulse width to make sure the system is working accurately as expected.

Keywords: Amplitude, Constant Pulse Width, Frequency Divider, Pulse Generator.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 3637

Authors: Eshwara K. Prasad, Raman R. Goud, Swadesh Kumar Singh, N. Sateesh

Abstract:

In the present work, forming limit diagrams and strain distribution profile diagrams for extra deep drawing steel at room and elevated temperatures have been determined experimentally by conducting stretch forming experiments by using designed and fabricated warm stretchforming tooling setup. With the help of forming Limit Diagrams (FLDs) and strain, distribution profile diagrams the formability of Extra Deep Drawing steel has been analyzed and co-related with mechanical properties like strain hardening COEFFICIENT (n) and normal anisotropy (r−). Mechanical properties of EDD steel from room temperature to 4500C were determined and discussed the impact of temperature on the properties like work hardening exponent (n) anisotropy (r-) and strength coefficient of the material. In addition, the fractured surfaces after stretching have undergone the some metallurgical investigations and attempt has been made to co-relate with the formability of EDD steel sheets. They are co-related and good agreement with FLDs at various temperatures.

Keywords: FLD, microhardness, strain distribution profile, stretch forming.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1811

Authors: Kripesh Adhikari, Hamid Bouchachia, Hammadi Nait-Charif

Abstract:

Falls are one of the major causes of injury and death among elderly people aged 65 and above. A support system to identify such kind of abnormal activities have become extremely important with the increase in ageing population. Pose estimation is a challenging task and to add more to this, it is even more challenging when pose estimations are performed on challenging poses that may occur during fall. Location of the body provides a clue where the person is at the time of fall. This paper presents a vision-based tracking strategy where available joints are grouped into three different feature points depending upon the section they are located in the body. The three feature points derived from different joints combinations represents the upper region or head region, mid-region or torso and lower region or leg region. Tracking is always challenging when a motion is involved. Hence the idea is to locate the regions in the body in every frame and consider it as the tracking strategy. Grouping these joints can be beneficial to achieve a stable region for tracking. The location of the body parts provides a crucial information to distinguish normal activities from falls.

Keywords: Fall detection, machine learning, deep learning, pose estimation, tracking.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 2087

Authors: JaeHwan Yang, Da-Woon Jeong, Seung-Young Kho, Dong-Kyu Kim

Abstract:

In terms of ITS, information on link characteristic is an essential factor for plan or operation. But in practical cases, not every link has installed sensors on it. The link that does not have data on it is called “Missing Link”. The purpose of this study is to impute data of these missing links. To get these data, this study applies the machine learning method. With the machine learning process, especially for the deep learning process, missing link data can be estimated from present link data. For deep learning process, this study uses “Recurrent Neural Network” to take time-series data of road. As input data, Dedicated Short-range Communications (DSRC) data of Dalgubul-daero of Daegu Metropolitan Area had been fed into the learning process. Neural Network structure has 17 links with present data as input, 2 hidden layers, for 1 missing link data. As a result, forecasted data of target link show about 94% of accuracy compared with actual data.

Keywords: Data Estimation, link data, machine learning, road network.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1471

Authors: Padmanabhan Balasubramanian, C. Hari Narayanan

Abstract:

This paper discusses a new, systematic approach to the synthesis of a NP-hard class of non-regenerative Boolean networks, described by FON[FOFF]={mi}[{Mi}], where for every mj[Mj]∈{mi}[{Mi}], there exists another mk[Mk]∈{mi}[{Mi}], such that their Hamming distance HD(mj, mk)=HD(Mj, Mk)=O(n), (where 'n' represents the number of distinct primary inputs). The method automatically ensures exact minimization for certain important selfdual functions with 2n-1 points in its one-set. The elements meant for grouping are determined from a newly proposed weighted incidence matrix. Then the binary value corresponding to the candidate pair is correlated with the proposed binary value matrix to enable direct synthesis. We recommend algebraic factorization operations as a post processing step to enable reduction in literal count. The algorithm can be implemented in any high level language and achieves best cost optimization for the problem dealt with, irrespective of the number of inputs. For other cases, the method is iterated to subsequently reduce it to a problem of O(n-1), O(n-2),.... and then solved. In addition, it leads to optimal results for problems exhibiting higher degree of adjacency, with a different interpretation of the heuristic, and the results are comparable with other methods. In terms of literal cost, at the technology independent stage, the circuits synthesized using our algorithm enabled net savings over AOI (AND-OR-Invert) logic, AND-EXOR logic (EXOR Sum-of- Products or ESOP forms) and AND-OR-EXOR logic by 45.57%, 41.78% and 41.78% respectively for the various problems. Circuit level simulations were performed for a wide variety of case studies at 3.3V and 2.5V supply to validate the performance of the proposed method and the quality of the resulting synthesized circuits at two different voltage corners. Power estimation was carried out for a 0.35micron TSMC CMOS process technology. In comparison with AOI logic, the proposed method enabled mean savings in power by 42.46%. With respect to AND-EXOR logic, the proposed method yielded power savings to the tune of 31.88%, while in comparison with AND-OR-EXOR level networks; average power savings of 33.23% was obtained.

Keywords: AOI logic, ESOP, AND-OR-EXOR, Incidencematrix, Hamming distance.

Procedia APA BibTeX Chicago EndNote Harvard JSON MLA RIS XML ISO 690 PDF Downloads 1498