Search results for: pooling

Authors: Lanchi Xie, Zhihui Li, Zhigang Li, Guiqiang Wang, Lei Xu, Yuwen Yan

Abstract:

Image-based facial features can be classified into category recognition features and individual recognition features. Current automated face recognition systems extract a specific feature vector of different dimensions from a facial image according to their pre-trained neural network. However, to improve the efficiency of parameter calculation, an algorithm generally reduces the image details by pooling. The operation will overlook the details concerned much by forensic experts. In our experiment, we adopted a variety of face recognition algorithms based on deep learning, compared a large number of naturally collected face images with the known data of the same person's frontal ID photos. Downscaling and manual handling were performed on the testing images. The results supported that the facial recognition algorithms based on deep learning detected structural and morphological information and rarely focused on specific markers such as stains and moles. Overall performance, distribution of genuine scores and impostor scores, and likelihood ratios were tested to evaluate the accuracy of biometric systems and forensic experts. Experiments showed that the biometric systems were skilled in distinguishing category features, and forensic experts were better at discovering the individual features of human faces. In the proposed approach, a fusion was performed at the score level. At the specified false accept rate, the framework achieved a lower false reject rate. This paper contributes to improving the interpretability of the objective method of facial comparison and provides a novel method for human-machine collaboration in this field.

Keywords: likelihood ratio, automated facial recognition, facial comparison, biometrics

Procedia PDF Downloads 99

Authors: Oumaima Khlifati, Khadija Baba

Abstract:

Pavement distress is the main factor responsible for the deterioration of road structure durability, damage vehicles, and driver comfort. Transportation agencies spend a high proportion of their funds on pavement monitoring and maintenance. The auscultation of pavement distress was based on the manual survey, which was extremely time consuming, labor intensive, and required domain expertise. Therefore, the automatic distress detection is needed to reduce the cost of manual inspection and avoid more serious damage by implementing the appropriate remediation actions at the right time. Inspired by recent deep learning applications, this paper proposes an algorithm for automatic road distress detection and classification using on the Deep Convolutional Neural Network (DCNN). In this study, the types of pavement distress are classified as transverse or longitudinal cracking, alligator, pothole, and intact pavement. The dataset used in this work is composed of public asphalt pavement images. In order to learn the structure of the different type of distress, the DCNN models are trained and tested as a multi-label classification task. In addition, to get the highest accuracy for our model, we adjust the structural optimization hyper parameters such as the number of convolutions and max pooling, filers, size of filters, loss functions, activation functions, and optimizer and fine-tuning hyper parameters that conclude batch size and learning rate. The optimization of the model is executed by checking all feasible combinations and selecting the best performing one. The model, after being optimized, performance metrics is calculated, which describe the training and validation accuracies, precision, recall, and F1 score.

Keywords: distress pavement, hyperparameters, automatic classification, deep learning

Procedia PDF Downloads 57

Authors: Yanling Li, Linying Ji, Zita Oravecz, Timothy R. Brick, Michael D. Hunter, Sy-Miin Chow

Abstract:

Assessing several individuals intensively over time yields intensive longitudinal data (ILD). Even though ILD provide rich information, they also bring other data analytic challenges. One of these is the increased occurrence of missingness with increased study length, possibly under non-ignorable missingness scenarios. Multiple imputation (MI) handles missing data by creating several imputed data sets, and pooling the estimation results across imputed data sets to yield final estimates for inferential purposes. In this article, we introduce dynr.mi(), a function in the R package, Dynamic Modeling in R (dynr). The package dynr provides a suite of fast and accessible functions for estimating and visualizing the results from fitting linear and nonlinear dynamic systems models in discrete as well as continuous time. By integrating the estimation functions in dynr and the MI procedures available from the R package, Multivariate Imputation by Chained Equations (MICE), the dynr.mi() routine is designed to handle possibly non-ignorable missingness in the dependent variables and/or covariates in a user-specified dynamic systems model via MI, with convergence diagnostic check. We utilized dynr.mi() to examine, in the context of a vector autoregressive model, the relationships among individuals’ ambulatory physiological measures, and self-report affect valence and arousal. The results from MI were compared to those from listwise deletion of entries with missingness in the covariates. When we determined the number of iterations based on the convergence diagnostics available from dynr.mi(), differences in the statistical significance of the covariate parameters were observed between the listwise deletion and MI approaches. These results underscore the importance of considering diagnostic information in the implementation of MI procedures.

Keywords: dynamic modeling, missing data, mobility, multiple imputation

Procedia PDF Downloads 141

Authors: Meron Okbandrias

Abstract:

The South African formal economy is rule-based economy, unlike most African and Asian markets. It has a highly developed financial market. In such a market, foreign migrants have dominated the small or spaza shops that service the poor. They are highly competitive and capture significant market share in South Africa. This paper analyses the factors that assisted the foreign migrants in having a competitive age. It does that by interviewing Somali, Bangladesh, and Ethiopian shop owners in Cape Town analysing the data through a narrative analysis. The paper also analyses the 2019 South African consumer report. The three migrant nationalities mentioned above dominate the spaza shop business and have significant distribution networks. The findings of the paper indicate that family, ethnic, and nationality based network, in that order of importance, form bases for a relationship-based business network that has trust as its mainstay. Therefore, this network ensures the pooling of resources and abiding by certain principles outside the South African rule-based system. The research identified practises like bulk buying within a community of traders, sharing information, buying from a within community distribution business, community based transportation system and providing seed capital for people from the community to start a business is all based on that relationship-based system. The consequences of not abiding by the rules of these networks are social and economic exclusion. In addition, these networks have their own commercial and social conflict resolution mechanisms aside from the South African justice system. Network theory and relationship based systems theory form the theoretical foundations of this paper.

Keywords: migrant, spaza shops, relationship-based system, South Africa

Procedia PDF Downloads 99

Authors: Robin Molinier

Abstract:

Industrial symbiosis (I.S) rely on two modes of cooperation that are infrastructure sharing and resource substitution to obtain economic and environmental benefits. The former consists in the intensification of use of an asset while the latter is based on the use of waste, fatal energy (and utilities) as alternatives to standard inputs. Both modes, in fact, rely on the shift from a business-as-usual functioning towards an alternative production system structure so that in a business point of view the distinction is not clear. In order to investigate the way those cooperation modes can be distinguished, we consider the stakeholders' interplay in the business model structure regarding their resources and requirements. For infrastructure sharing (following economic engineering literature) the cost function of capacity induces economies of scale so that demand pooling reduces global expanses. Grassroot investment sizing decision and the ex-post pricing strongly depends on the design optimization phase for capacity sizing whereas ex-post operational cost sharing minimizing budgets are less dependent upon production rates. Value is then mainly design driven. For resource substitution, synergies value stems from availability and is at risk regarding both supplier and user load profiles and market prices of the standard input. Baseline input purchasing cost reduction is thus more driven by the operational phase of the symbiosis and must be analyzed within the whole sourcing policy (including diversification strategies and expensive back-up replacement). Moreover, while resource substitution involves a chain of intermediate processors to match quality requirements, the infrastructure model relies on a single operator whose competencies allow to produce non-rival goods. Transaction costs appear higher in resource substitution synergies due to the high level of customization which induces asset specificity, and non-homogeneity following transaction costs economics arguments.

Keywords: business model, capacity, sourcing, synergies

Procedia PDF Downloads 148

Authors: Veerpal Kaur

Abstract:

The present paper aims at investigating the role of dynamic externalities of agglomeration in the regional productivity growth of manufacturing sector in India and China. Taking 2-digit level manufacturing sector data of states and provinces of India and China respectively for the period of 1998-99 to 2011-12, this paper examines the effect of dynamic externalities namely – Marshall-Arrow-Romer (MAR) specialization externalities, Jacobs’s diversity externalities, and Porter’s competition externalities on regional total factor productivity growth (TFPG) of manufacturing sector in both economies. Regressions have been carried on pooled data for all 2-digit manufacturing industries for India and China separately. The estimation of Panel has been based on a fixed effect by sector model. The results of econometric exercise show that labour-intensive industries in Indian regional manufacturing benefit from diversity externalities and capital intensive industries gain more from specialization in terms of TFPG. In China, diversity externalities and competition externalities hold better prospectus for regional TFPG in both labour intensive and capital intensive industries. But if we look at results for coastal and non-coastal region separately, specialization tends to assert a positive effect on TFPG in coastal regions whereas it has a negative effect on TFPG of coastal regions. Competition externalities put a negative effect on TFPG of non-coastal regions whereas it has a positive effect on TFPG of coastal regions. Diversity externalities made a positive contribution to TFPG in both coastal and non-coastal regions. So the results of the study postulate that the importance of dynamic externalities should not be examined by pooling all industries and all regions together. This could hold differential implications for region specific and industry-specific policy formulation. Other important variables explaining regional level TFPG in both India and China have been the availability of infrastructure, level of competitiveness, foreign direct investment, exports and geographical location of the region (especially in China).

Keywords: China, dynamic externalities, India, manufacturing, productivity

Procedia PDF Downloads 100

Authors: Monica Gonzalez Machorro

Abstract:

Dementia is hard to diagnose because of the lack of early physical symptoms. Early dementia recognition is key to improving the living condition of patients. Speech technology is considered a valuable biomarker for this challenge. Recent works have utilized conventional acoustic features and machine learning methods to detect dementia in speech. BERT-like classifiers have reported the most promising performance. One constraint, nonetheless, is that these studies are either based on human transcripts or on transcripts produced by automatic speech recognition (ASR) systems. This research contribution is to explore a method that does not require transcriptions to detect early Alzheimer’s disease (AD) and mild cognitive impairment (MCI). This is achieved by fine-tuning a pre-trained ASR model for the downstream early AD and MCI tasks. To do so, a subset of the thoroughly studied Pitt Corpus is customized. The subset is balanced for class, age, and gender. Data processing also involves cropping the samples into 10-second segments. For comparison purposes, a baseline model is defined by training and testing a Random Forest with 20 extracted acoustic features using the librosa library implemented in Python. These are: zero-crossing rate, MFCCs, spectral bandwidth, spectral centroid, root mean square, and short-time Fourier transform. The baseline model achieved a 58% accuracy. To fine-tune HuBERT as a classifier, an average pooling strategy is employed to merge the 3D representations from audio into 2D representations, and a linear layer is added. The pre-trained model used is ‘hubert-large-ls960-ft’. Empirically, the number of epochs selected is 5, and the batch size defined is 1. Experiments show that our proposed method reaches a 69% balanced accuracy. This suggests that the linguistic and speech information encoded in the self-supervised ASR-based model is able to learn acoustic cues of AD and MCI.

Keywords: automatic speech recognition, early Alzheimer’s recognition, mild cognitive impairment, speech impairment

Procedia PDF Downloads 97

Authors: Tomohiko Utsuki

Abstract:

Currently, brain resuscitation becomes increasingly important due to revising various clinical guidelines pertinent to emergency care. In brain resuscitation, the control of brain temperature (BT), intracranial pressure (ICP), and cerebral blood flow (CBF) is required for stabilizing physiological state of brain, and is described as the essential treatment points in many guidelines of disorder and/or disease such as brain injury, stroke, and encephalopathy. Thus, an integrated control system of BT, ICP, and CBF will greatly contribute to alleviating the burden on medical staff and improving treatment effect in brain resuscitation. In order to develop such a control system, models related to BT, ICP, and CBF are required for control simulation, because trial and error experiments using patients are not ethically allowed. A static model of cerebral blood circulation from intracranial arteries and vertebral artery to jugular veins has already constructed and verified. However, it is impossible to represent the pooling of blood in blood vessels, which is one cause of cerebral hypertension in this model. And, it is also impossible to represent the pulsing motion of blood vessels caused by blood pressure change which can have an affect on the change of cerebral tissue pressure. Thus, a dynamic model of cerebral blood circulation is constructed in consideration of the elasticity of the blood vessel and the inertia of the blood vessel wall. The constructed dynamic model was numerically analyzed using the normal data, in which each arterial blood flow in cerebral blood circulation, the distribution of blood pressure in the Circle of Willis, and the change of blood pressure along blood flow were calculated for verifying against physiological knowledge. As the result, because each calculated numerical value falling within the generally known normal range, this model has no problem in representing at least the normal physiological state of the brain. It is the next task to verify the accuracy of the present model in the case of disease or disorder. Currently, the construction of a migration model of extracellular fluid and a model of heat transfer in cerebral tissue are in progress for making them parts of an integrated model of brain physiological state, which is necessary for developing an future integrated control system of BT, ICP and CBF. The present model is applicable to constructing the integrated model representing at least the normal condition of brain physiological state by uniting with such models.

Keywords: dynamic model, cerebral blood circulation, brain resuscitation, automatic control

Procedia PDF Downloads 132

Authors: Aparna Soni

Abstract:

The much-celebrated foretold story of Indian city engines driving the growth of India has been scrutinized to have serious consequences. A wide spectrum of scholarship has brought to light the un-equalizing effects and the need to adopt a rights-based approach to development planning in India. Notably, these concepts and discourses ubiquitously entail the study of land struggles in the making of Urban. In fact, the very progression of the primitive accumulation theory to accumulation by dispossession, followed by ‘dispossession without development,’ thereafter Development without dispossession and now as Dispossession by financialization noticeably the last three developing in a span of mere three decades, is evidence enough to trace the centrality and evolving role of land in the making of urban India. India, in the last decade, has seen its regional governments actively experimenting with alternative models of land assembly (Amaravati and Delhi land pooling models, the loudly advertised ones). These are publicized as a replacement to the presumably cost and time antagonistic, prone to litigation land acquisition act of 2013. It has been observed that most of the literature treats these models as a generic large bracket of land expropriation and do not, in particular, try to differentially analyse to granularly find a pattern in these alternatives. To cater to this gap, this research comparatively studies these alternative land, assembly models. It categorises them based on their basic architecture, spatial and sectoral application, and governance frameworks. It is found that these alternatives are ad-hoc and fragmented pieces of legislation. These are fit for profit models commodifying land to ease its access by the private sector for real estate led growth. The research augments the literature on the privatization of land use planning in India. Further, it attempts to discuss the increasing role a landowner is expected to play in the future and suggests a way forward to safeguard them from market risks. The study involves a thematic analysis of the policy elements contained in legislative/policy documents, notifications, office orders. The study also derives from the various widely circulated print media information. With the present field-visit limitations, the study relies on documents accessed open-source in the public domain.

Keywords: commodification, dispossession, land acquisition, landowner

Procedia PDF Downloads 132

Authors: Yanjun Chen, Chuanping Hu, Jie Shao, Lin Mei, Chongyang Zhang

Abstract:

Traditional image classification is based on encoding scheme (e.g. Fisher Vector, Vector of Locally Aggregated Descriptor) with low-level image features (e.g. SIFT, HoG). Compared to these low-level local features, deep convolutional features obtained at the mid-level layer of convolutional neural networks (CNN) have richer information but lack of geometric invariance. For scene classification, there are scattered objects with different size, category, layout, number and so on. It is crucial to find the distinctive objects in scene as well as their co-occurrence relationship. In this paper, we propose a method to take advantage of both deep convolutional features and the traditional encoding scheme while taking object-centric and scene-centric information into consideration. First, to exploit the object-centric and scene-centric information, two CNNs that trained on ImageNet and Places dataset separately are used as the pre-trained models to extract deep convolutional features at multiple scales. This produces dense local activations. By analyzing the performance of different CNNs at multiple scales, it is found that each CNN works better in different scale ranges. A scale-wise CNN adaption is reasonable since objects in scene are at its own specific scale. Second, a fisher kernel is applied to aggregate a global representation at each scale and then to merge into a single vector by using a post-processing method called scale-wise normalization. The essence of Fisher Vector lies on the accumulation of the first and second order differences. Hence, the scale-wise normalization followed by average pooling would balance the influence of each scale since different amount of features are extracted. Third, the Fisher vector representation based on the deep convolutional features is followed by a linear Supported Vector Machine, which is a simple yet efficient way to classify the scene categories. Experimental results show that the scale-specific feature extraction and normalization with CNNs trained on object-centric and scene-centric datasets can boost the results from 74.03% up to 79.43% on MIT Indoor67 when only two scales are used (compared to results at single scale). The result is comparable to state-of-art performance which proves that the representation can be applied to other visual recognition tasks.

Keywords: deep convolutional features, Fisher Vector, multiple scales, scale-specific normalization

Procedia PDF Downloads 301

Authors: Daniel Nigusse Tollosa, Erica James, Alexis Hurre, Meredith Tavener

Abstract:

Introduction: Lifestyle behaviours such as healthy diet, regular physical activity and maintaining a healthy weight are essential for cancer survivors to improve the quality of life and longevity. However, there is no study that synthesis cancer survivor’s adherence to healthy lifestyle recommendations. The purpose of this review was to collate existing data on the prevalence of adherence to healthy behaviours and produce the pooled estimate among adult cancer survivors. Method: Multiple databases (Embase, Medline, Scopus, Web of Science and Google Scholar) were searched for relevant articles published since 2007, reporting cancer survivors adherence to more than two lifestyle behaviours based on the WCRF/AICR recommendations. The pooled prevalence of adherence to single and multiple behaviours (operationalized as adherence to more than 75% (3/4) of health behaviours included in a particular study) was calculated using a random effects model. Subgroup analysis adherence to multiple behaviours was undertaken corresponding to the mean survival years and year of publication. Results: A total of 3322 articles were generated through our search strategies. Of these, 51 studies matched our inclusion criteria, which presenting data from 2,620,586 adult cancer survivors. The highest prevalence of adherence was observed for smoking (pooled estimate: 87%, 95% CI: 85%, 88%) and alcohol intake (pooled estimate 83%, 95% CI: 81%, 86%), and the lowest was for fiber intake (pooled estimate: 31%, 95% CI: 21%, 40%). Thirteen studies were reported the proportion of cancer survivors (all used a simple summative index method) to multiple healthy behaviours, whereby the prevalence of adherence was ranged from 7% to 40% (pooled estimate 23%, 95% CI: 17% to 30%). Subgroup analysis suggest that short-term survivors ( < 5 years survival time) had relatively a better adherence to multiple behaviours (pooled estimate: 31%, 95% CI: 27%, 35%) than long-term ( > 5 years survival time) cancer survivors (pooled estimate: 25%, 95% CI: 14%, 36%). Pooling of estimates according to the year of publication (since 2007) also suggests an increasing trend of adherence to multiple behaviours over time. Conclusion: Overall, the adherence to multiple lifestyle behaviors was poor (not satisfactory), and relatively, it is a major concern for long-term than the short-term cancer survivor. Cancer survivors need to obey with healthy lifestyle recommendations related to physical activity, fruit and vegetable, fiber, red/processed meat and sodium intake.

Keywords: adherence, lifestyle behaviours, cancer survivors, WCRF/AICR

Procedia PDF Downloads 157

Authors: Oliver Chunho Ma, Tszying Chang

Abstract:

Background: Traditional Chinese Medicine (TCM) has been widely used in the treatment of eczema. However, there is currently a lack of comprehensive research on the overall effectiveness of TCM in treating eczema, particularly using the Eczema Area and Severity Index (EASI) score as an evaluation tool. Meta-analysis can integrate the results of multiple studies to provide more convincing evidence. Objective: To conduct a systematic review and meta-analysis based on the EASI score to evaluate the overall effectiveness of TCM in the treatment of eczema. Specifically, the study will review and analyze published clinical studies that investigate TCM treatments for eczema and use the EASI score as an outcome measure, comparing the differences in improving the severity of eczema between TCM and other treatment modalities, such as conventional Western medicine treatments. Methods: Relevant studies, including randomized controlled trials (RCTs) and non-randomized controlled trials, that involve TCM treatment for eczema and use the EASI score as an outcome measure will be searched in medical literature databases such as PubMed, CNKI, etc. Relevant data will be extracted from the selected studies, including study design, sample size, treatment methods, improvement in EASI score, etc. The methodological quality and risk of bias of the included studies will be assessed using appropriate evaluation tools (such as the Cochrane Handbook). The results of the selected studies will be statistically analyzed, including pooling effect sizes (such as standardized mean differences, relative risks, etc.), subgroup analysis (e.g., different TCM syndromes, different treatment modalities), and sensitivity analysis (e.g., excluding low-quality studies). Based on the results of the statistical analysis and quality assessment, the overall effectiveness of TCM in improving the severity of eczema will be interpreted. Expected outcomes: By integrating the results of multiple studies, we expect to provide more convincing evidence regarding the specific effects of TCM in improving the severity of eczema. Additionally, subgroup analysis and sensitivity analysis can further elucidate whether the effectiveness of TCM treatment is influenced by different factors. Besides, we will compare the results of the meta-analysis with the clinical data from our clinic. For both the clinical data and the meta-analysis results, we will perform descriptive statistics such as means, standard deviations, percentages, etc. and compare the differences between the two using statistical tests such as independent samples t-test or non-parametric tests to assess the statistical differences between them.

Keywords: Eczema, traditional Chinese medicine, EASI, systematic review, meta-analysis

Procedia PDF Downloads 26

Authors: Tassadit Issaadi Hamitouche, Nicolas Gillard, Jean Petit, Valerie Lavaste, Celine Mayousse

Abstract:

Water is a fundamental resource for the industry. It is taken from the environment either from municipal distribution networks or from various natural water sources such as the sea, ocean, rivers, aquifers, etc. Once used, water is discharged into the environment, reprocessed at the plant or treatment plants. These withdrawals and discharges have a direct impact on natural water resources. These impacts can apply to the quantity of water available, the quality of the water used, or to impacts that are more complex to measure and less direct, such as the health of the population downstream from the watercourse, for example. Based on the analysis of data (meteorological, river characteristics, physicochemical substances), we wish to predict water stress episodes and anticipate prefectoral decrees, which can impact the performance of plants and propose improvement solutions, help industrialists in their choice of location for a new plant, visualize possible interactions between companies to optimize exchanges and encourage the pooling of water treatment solutions, and set up circular economies around the issue of water. The development of a system for the collection, processing, and use of data related to water resources requires the functional constraints specific to the latter to be made explicit. Thus the system will have to be able to store a large amount of data from sensors (which is the main type of data in plants and their environment). In addition, manufacturers need to have 'near-real-time' processing of information in order to be able to make the best decisions (to be rapidly notified of an event that would have a significant impact on water resources). Finally, the visualization of data must be adapted to its temporal and geographical dimensions. In this study, we set up an infrastructure centered on the TICK application stack (for Telegraf, InfluxDB, Chronograf, and Kapacitor), which is a set of loosely coupled but tightly integrated open source projects designed to manage huge amounts of time-stamped information. The software architecture is coupled with the cross-industry standard process for data mining (CRISP-DM) data mining methodology. The robust architecture and the methodology used have demonstrated their effectiveness on the study case of learning the level of a river with a 7-day horizon. The management of water and the activities within the plants -which depend on this resource- should be considerably improved thanks, on the one hand, to the learning that allows the anticipation of periods of water stress, and on the other hand, to the information system that is able to warn decision-makers with alerts created from the formalization of prefectoral decrees.

Keywords: data mining, industry, machine Learning, shortage, water resources

Procedia PDF Downloads 92

Authors: Varun Agarwal

Abstract:

Introduction: With the advent of personalized medicine, histopathological review of whole slide images (WSIs) for cancer diagnosis presents an exceedingly time-consuming, complex task. Specifically, detecting metastatic regions in WSIs of sentinel lymph node biopsies necessitates a full-scanned, holistic evaluation of the image. Thus, digital pathology, low-level image manipulation algorithms, and machine learning provide significant advancements in improving the efficiency and accuracy of WSI analysis. Using Camelyon16 data, this paper proposes a deep learning pipeline to automate and ameliorate breast cancer metastasis localization and WSI classification. Methodology: The model broadly follows five stages -region of interest detection, WSI partitioning into image tiles, convolutional neural network (CNN) image-segment classifications, probabilistic mapping of tumor localizations, and further processing for whole WSI classification. Transfer learning is applied to the task, with the implementation of Inception-ResNetV2 - an effective CNN classifier that uses residual connections to enhance feature representation, adding convolved outputs in the inception unit to the proceeding input data. Moreover, in order to augment the performance of the transfer learning CNN, a stack of convolutional denoising autoencoders (CDAE) is applied to produce embeddings that enrich image representation. Through a saliency-detection algorithm, visual training segments are generated, which are then processed through a denoising autoencoder -primarily consisting of convolutional, leaky rectified linear unit, and batch normalization layers- and subsequently a contrast-normalization function. A spatial pyramid pooling algorithm extracts the key features from the processed image, creating a viable feature map for the CNN that minimizes spatial resolution and noise. Results and Conclusion: The simplified and effective architecture of the fine-tuned transfer learning Inception-ResNetV2 network enhanced with the CDAE stack yields state of the art performance in WSI classification and tumor localization, achieving AUC scores of 0.947 and 0.753, respectively. The convolutional feature retention and compilation with the residual connections to inception units synergized with the input denoising algorithm enable the pipeline to serve as an effective, efficient tool in the histopathological review of WSIs.

Keywords: breast cancer, convolutional neural networks, metastasis mapping, whole slide images

Procedia PDF Downloads 105

Authors: Manaranjan Pradhan, Shailaja Grover, U. Dinesh Kumar

Abstract:

Farming community in India, as well as other parts of the world, is one of the highly stressed communities due to reasons such as increasing input costs (cost of seeds, fertilizers, pesticide), droughts, reduced revenue leading to farmer suicides. Lack of integrated farm advisory system in India adds to the farmers problems. Farmers need right information during the early stages of crop’s lifecycle to prevent damage and loss in revenue. In this paper, we use deep learning techniques to develop an early warning system for detection of crop diseases using images taken by farmers using their smart phone. The research work leads to building a smart assistant using analytics and big data which could help the farmers with early diagnosis of the crop diseases and corrective actions. The classical approach for crop disease management has been to identify diseases at crop level. Recently, ImageNet Classification using the convolutional neural network (CNN) has been successfully used to identify diseases at individual plant level. Our model uses convolution filters, max pooling, dense layers and dropouts (to avoid overfitting). The models are built for binary classification (healthy or not healthy) and multi class classification (identifying which disease). Transfer learning is used to modify the weights of parameters learnt through ImageNet dataset and apply them on crop diseases, which reduces number of epochs to learn. One shot learning is used to learn from very few images, while data augmentation techniques are used to improve accuracy with images taken from farms by using techniques such as rotation, zoom, shift and blurred images. Models built using combination of these techniques are more robust for deploying in the real world. Our model is validated using tomato crop. In India, tomato is affected by 10 different diseases. Our model achieves an accuracy of more than 95% in correctly classifying the diseases. The main contribution of our research is to create a personal assistant for farmers for managing plant disease, although the model was validated using tomato crop, it can be easily extended to other crops. The advancement of technology in computing and availability of large data has made possible the success of deep learning applications in computer vision, natural language processing, image recognition, etc. With these robust models and huge smartphone penetration, feasibility of implementation of these models is high resulting in timely advise to the farmers and thus increasing the farmers' income and reducing the input costs.

Keywords: analytics in agriculture, CNN, crop disease detection, data augmentation, image recognition, one shot learning, transfer learning

Procedia PDF Downloads 95

Authors: Shunhong Qi, Hui Tian

Abstract:

Triggered off by the forthcoming 2022 Winter Olympics, ski centers are blossoming in China, the number being 742 in 2018. Although the number of skier visits of ski resorts soared to 19.7 million in 2018, one-time skiers account for a considerable portion therein. In light of the extremely low return rates and skiing penetration level (0.5%) of leisure skiing in China, this study proposes and tests a leisure ski service performance framework which assesses the ski resorts’ service quality, skier satisfaction, as well as their impact on skiers’ behavioral intentions, with an aim to assess the success of ski resorts and provide suggestions for improvement. Three self-administered surveys and 16 interviews were conducted upon a convenience sample of leisure skiers in two major ski destinations within two hours’ drive from Beijing – Nanshan and Jundushan ski resorts. Of the 680 questionnaires distributed, 416 usable copies were returned, the response rate being 61.2%. The questionnaire used for the study was developed based on the existing literature of 'push' factors of skiers (intrinsic desire) and 'pull' factors (attractiveness of a destination), as well as leisure sport satisfaction. The scale comprises four parts: skiers’ demographic profiles, their perceived service quality (including ski resorts’ infrastructure, expense, safety and comfort, convenience, daily needs support, skill development support, and accessibility), their overall levels of satisfaction (satisfaction with the service and the experience), and their behavioral intentions (including loyalty, future visitation and greater tolerance of price increases). Skiers’ demographic profiles show that among the 220 males and 196 females in the survey, a vast majority of the skiers are age 17-39 (87.2%). 64.7% are not married, and nearly half (48.3%) of the skiers have a monthly family income exceeding 10,000 yuan (USD 1,424), and 80% are beginners or intermediate skiers. The regression examining the influence of service quality on skier satisfaction reveals that service quality accounts for 44.4% of the variance in skier satisfaction, the variables of safety and comfort, expense, skill development support, and accessibility contributing significantly in descending order. Another regression analyzing the influence of service quality as well as skier satisfaction on their behavioral intentions shows that service quality and skier satisfaction account for 39.1% of the variance in skiers’ behavioral intentions, and the significant predictors are skier satisfaction, safety and comfort, expense, and accessibility, in descending order, though a comparison between groups also indicates that for expert skiers, the significant variables are skier satisfaction, skill development support, safety, and comfort. Suggestions are thus made for ski resorts and other stakeholders to improve skier satisfaction and increase visitation: developing diversified ski courses to meet the demands of skiers of different skiing skills and to reduce crowding, adopting enough chairlifts and magic carpets, reinforcing safety measures and medical force; further exploring their various resources and lower the skiing expense on ski pass, equipment renting, accommodation and dining; adding more bus lines and/or develop platforms for skiers’ car-pooling, and offering diversified skiing activities with local flavors for better entertainment.

Keywords: behavioral intentions, leisure skiing, service quality, skier satisfaction

Procedia PDF Downloads 67

Authors: Piotr A. Kowalski, Kasper Sapala, Wiktor Warchalowski

Abstract:

PM10 is a suspended dust that primarily has a negative effect on the respiratory system. PM10 is responsible for attacks of coughing and wheezing, asthma or acute, violent bronchitis. Indirectly, PM10 also negatively affects the rest of the body, including increasing the risk of heart attack and stroke. Unfortunately, Poland is a country that cannot boast of good air quality, in particular, due to large PM concentration levels. Therefore, based on the dense network of Airly sensors, it was decided to deal with the problem of prediction of suspended particulate matter concentration. Due to the very complicated nature of this issue, the Machine Learning approach was used. For this purpose, Convolution Neural Network (CNN) neural networks have been adopted, these currently being the leading information processing methods in the field of computational intelligence. The aim of this research is to show the influence of particular CNN network parameters on the quality of the obtained forecast. The forecast itself is made on the basis of parameters measured by Airly sensors and is carried out for the subsequent day, hour after hour. The evaluation of learning process for the investigated models was mostly based upon the mean square error criterion; however, during the model validation, a number of other methods of quantitative evaluation were taken into account. The presented model of pollution prediction has been verified by way of real weather and air pollution data taken from the Airly sensor network. The dense and distributed network of Airly measurement devices enables access to current and archival data on air pollution, temperature, suspended particulate matter PM1.0, PM2.5, and PM10, CAQI levels, as well as atmospheric pressure and air humidity. In this investigation, PM2.5, and PM10, temperature and wind information, as well as external forecasts of temperature and wind for next 24h served as inputted data. Due to the specificity of the CNN type network, this data is transformed into tensors and then processed. This network consists of an input layer, an output layer, and many hidden layers. In the hidden layers, convolutional and pooling operations are performed. The output of this system is a vector containing 24 elements that contain prediction of PM10 concentration for the upcoming 24 hour period. Over 1000 models based on CNN methodology were tested during the study. During the research, several were selected out that give the best results, and then a comparison was made with the other models based on linear regression. The numerical tests carried out fully confirmed the positive properties of the presented method. These were carried out using real ‘big’ data. Models based on the CNN technique allow prediction of PM10 dust concentration with a much smaller mean square error than currently used methods based on linear regression. What's more, the use of neural networks increased Pearson's correlation coefficient (R²) by about 5 percent compared to the linear model. During the simulation, the R² coefficient was 0.92, 0.76, 0.75, 0.73, and 0.73 for 1st, 6th, 12th, 18th, and 24th hour of prediction respectively.

Keywords: air pollution prediction (forecasting), machine learning, regression task, convolution neural networks

Procedia PDF Downloads 100

Authors: Fathonah Siti, Ardiana Irma

Abstract:

Indonesia is currently on its track to achieve Universal Health Coverage (UHC) by 2019. The program aims to address issues on disintegration in the implementation and coverage of various health insurance schemes and fragmented fund pooling. Family planning service is covered as one of benefit packages under preventive care. However, little has been done to examine how family planning program are appropriately managed across levels of governments and how family planning services are delivered to the end user. The study is performed through focus group discussion to related policy makers and selected programmers at central and district levels. The study is also benefited from relevant studies on family planning in the UHC scheme and other supporting data. The study carefully investigates some programmatic implications when family planning is integrated in the UHC program encompassing the need to recalculate contraceptive logistics for beneficiaries (eligible couple); policy reformulation for contraceptive service provision including supply chain management; establishment of family planning standard of procedure; and a call to update Management Information System. The study confirms that there is a significant increase in the numbers of contraceptive commodities needs to be procured by the government. Holding an assumption that contraceptive prevalence rate and commodities cost will be as expected increasing at 0.5% annually, the government need to allocate almost IDR 5 billion by 2019, excluded fee for service. The government shifts its focus to maintain eligible health facilities under National Population and Family Planning Board networks. By 2019, the government has set strategies to anticipate the provision of family planning services to 45.340 health facilities distributed in 514 districts and 7 thousand sub districts. Clear division of authorities has been established among levels of governments. Three models of contraceptive supply planning have been developed and currently in the process of being institutionalized. Pre service training for family planning services has been piloted in 10 prominent universities. The position of private midwives has been appreciated as part of the system. To ensure the implementation of quality and health expenditure control, family planning standard has been established as a reference to determine set of services required to deliver to the clients properly and types of health facilities to conduct particular family planning services. Recognition to individual status of program participation has been acknowledged in the Family Enumeration since 2015. The data is precisely recorded by name by address for each family and its members. It supplies valuable information to 15.131 Family Planning Field Workers (FPFWs) to provide information and education related to family planning in an attempt to generate demand and maintain the participation of family planning acceptors who are program beneficiaries. Despite overwhelming efforts described above, some obstacles remain. The program experiences poor socialization and yet removes geographical barriers for those living in remote areas. Family planning services provided for this sub population conducted outside the scheme as a complement strategy. However, UHC program has brought remarkable improvement in access and quality of family planning services.

Keywords: beneficiary, family planning services, national population and family planning board, universal health coverage

Procedia PDF Downloads 154