Search results for: text segmentation

Authors: Misrak Assefa

Abstract:

The Amharic language is the second most widely spoken Semitic language in the world. There are several new overloaded on the web. Searching some useful documents from the web on a specific topic, which is written in the Amharic language, is a challenging task. Hence, document categorization is required for managing and filtering important information. In the classification of Amharic text news, there is still a gap in the domain of information that needs to be launch. This study attempts to design an automatic Amharic news classification using a supervised learning mechanism on four un-touch classes. To achieve this research, 4,182 news articles were used. Naive Bayes (NB) and Decision tree (j48) algorithms were used to classify the given Amharic dataset. In this paper, k-fold cross-validation is used to estimate the accuracy of the classifier. As a result, it shows those algorithms can be applicable in Amharic news categorization. The best average accuracy result is achieved by j48 decision tree and naïve Bayes is 95.2345 %, and 94.6245 % respectively using three categories. This research indicated that a typical decision tree algorithm is more applicable to Amharic news categorization.

Keywords: text categorization, supervised machine learning, naive Bayes, decision tree

Procedia PDF Downloads 211

Authors: U. S. N. Raju, Kothuri Sai Kiran, Meena G. Kamal, Vinay Nikhil Pabba, Suresh Kanaparthi

Abstract:

There is huge amount of lecture video data available for public use, and many more lecture videos are being created and uploaded every day. Searching for videos on required topics from this huge database is a challenging task. Therefore, an efficient method for video retrieval is needed. An approach for automated video indexing and video search in large lecture video archives is presented. As the amount of video lecture data is huge, it is very inefficient to do the processing in a centralized computation framework. Hence, Hadoop Framework for distributed computing for Big Video Data is used. First, step in the process is automatic video segmentation and key-frame detection to offer a visual guideline for the video content navigation. In the next step, we extract textual metadata by applying video Optical Character Recognition (OCR) technology on key-frames. The OCR and detected slide text line types are adopted for keyword extraction, by which both video- and segment-level keywords are extracted for content-based video browsing and search. The performance of the indexing process can be improved for a large database by using distributed computing on Hadoop framework.

Keywords: video lectures, big video data, video retrieval, hadoop

Procedia PDF Downloads 537

Authors: Shaima Almalhan, Lubna Shukri, Miriam Talal, Safaa Teskieh

Abstract:

Since artificial intelligence is a rapidly evolving topic that has had a significant impact on technical growth and innovation, this paper examines people's awareness, use, and engagement with the Google Translate application. To see how familiar aware users are with the app and its features, quantitative and qualitative research was conducted. The findings revealed that consumers have a high level of confidence in the application and how far people they benefit from this sort of innovation and how convenient it makes communication.

Keywords: artificial intelligence, google translate, speech recognition, language translation, camera translation, speech to text, text to speech

Procedia PDF Downloads 155

Authors: Anass Hsissi, Hakim Allali, Abdelmajid Hajami

Abstract:

This article presents the results of a doctoral research which aims to integrate metacognitive dimension in the design of human learning computing environments (ILE). We conducted a detailed study on the relationship between metacognitive processes and learning, specifically their positive impact on the performance of learners in the area of reading comprehension. Our contribution is to implement methods, using an intelligent agent based on BDI paradigm to ensure intelligent and reliable support for low readers, in order to encourage regulation and a conscious and rational use of their metacognitive abilities.

Keywords: metacognition, text comprehension EIAH, autoregulation, BDI agent

Procedia PDF Downloads 323

Authors: Lamees Nasser, Yago Diez, Robert Martí, Joan Martí, Ibrahim Sadek

Abstract:

Automated 3D breast ultrasound (ABUS) screening is a novel modality in medical imaging because of its common characteristics shared with other ultrasound modalities in addition to the three orthogonal planes (i.e., axial, sagittal, and coronal) that are useful in analysis of tumors. In the literature, few automatic approaches exist for typical tasks such as segmentation or registration. In this work, we deal with two problems concerning ABUS images: nipple and rib detection. Nipple and ribs are the most visible and salient features in ABUS images. Determining the nipple position plays a key role in some applications for example evaluation of registration results or lesion follow-up. We present a nipple detection algorithm based on color and shape of the nipple, besides an automatic approach to detect the ribs. In point of fact, rib detection is considered as one of the main stages in chest wall segmentation. This approach consists of four steps. First, images are normalized in order to minimize the intensity variability for a given set of regions within the same image or a set of images. Second, the normalized images are smoothed by using anisotropic diffusion filter. Next, the ribs are detected in each slice by analyzing the eigenvalues of the 3D Hessian matrix. Finally, a breast mask and a probability map of regions detected as ribs are used to remove false positives (FP). Qualitative and quantitative evaluation obtained from a total of 22 cases is performed. For all cases, the average and standard deviation of the root mean square error (RMSE) between manually annotated points placed on the rib surface and detected points on rib borders are 15.1188 mm and 14.7184 mm respectively.

Keywords: Automated 3D Breast Ultrasound, Eigenvalues of Hessian matrix, Nipple detection, Rib detection

Procedia PDF Downloads 331

Authors: B. Kishore Kumar, Pogula Rakesh, T. Kishore Kumar

Abstract:

The music analysis is a part of the audio content analysis used to analyze the music by using the different features of audio signal. In music analysis, the first step is to divide the music signal to different sections based on the feature profiles of the music signal. In this paper, we present a music segmentation technique that will effectively segmentize the signal and thresholding technique to remove silence from the percussive sounds produced by percussive instruments, which uses two features of music, namely signal energy and spectral centroid. The proposed method impose thresholds on both the features which will vary depends on the music signal. Depends on the threshold, silence part is removed and the segmentation is done. The effectiveness of the proposed method is analyzed using MATLAB.

Keywords: percussive sounds, spectral centroid, spectral energy, silence removal, feature extraction

Procedia PDF Downloads 594

Authors: Suparman

Abstract:

Piecewise polynomial regression model is very flexible model for modeling the data. If the piecewise polynomial regression model is matched against the data, its parameters are not generally known. This paper studies the parameter estimation problem of piecewise polynomial regression model. The method which is used to estimate the parameters of the piecewise polynomial regression model is Bayesian method. Unfortunately, the Bayes estimator cannot be found analytically. Reversible jump MCMC algorithm is proposed to solve this problem. Reversible jump MCMC algorithm generates the Markov chain that converges to the limit distribution of the posterior distribution of piecewise polynomial regression model parameter. The resulting Markov chain is used to calculate the Bayes estimator for the parameters of piecewise polynomial regression model.

Keywords: piecewise regression, bayesian, reversible jump MCMC, segmentation

Procedia PDF Downloads 373

Authors: Joseph M. Rich, Vinay A. Duddalwar, Assad A. Oberai

Abstract:

Prostate adenocarcinoma is the most common cancer in males, with osseous metastases as the commonest site of metastatic prostate carcinoma (mPC). Treatment monitoring is based on the evaluation and characterization of lesions on multiple imaging studies, including Computed Tomography (CT). Monitoring of the osseous disease burden, including follow-up of lesions and identification and characterization of new lesions, is a laborious task for radiologists. Deep learning algorithms are increasingly used to perform tasks such as identification and segmentation for osseous metastatic disease and provide accurate information regarding metastatic burden. Here, nnUNet was used to produce a model which can segment CT scan images of prostate adenocarcinoma vertebral bone metastatic lesions. nnUNet is an open-source Python package that adds optimizations to deep learning-based UNet architecture but has not been extensively combined with transfer learning techniques due to the absence of a readily available functionality of this method. The IRB-approved study data set includes imaging studies from patients with mPC who were enrolled in clinical trials at the University of Southern California (USC) Health Science Campus and Los Angeles County (LAC)/USC medical center. Manual segmentation of metastatic lesions was completed by an expert radiologist Dr. Vinay Duddalwar (20+ years in radiology and oncologic imaging), to serve as ground truths for the automated segmentation. Despite nnUNet’s success on some medical segmentation tasks, it only produced an average Dice Similarity Coefficient (DSC) of 0.31 on the USC dataset. DSC results fell in a bimodal distribution, with most scores falling either over 0.66 (reasonably accurate) or at 0 (no lesion detected). Applying more aggressive data augmentation techniques dropped the DSC to 0.15, and reducing the number of epochs reduced the DSC to below 0.1. Datasets have been identified for transfer learning, which involve balancing between size and similarity of the dataset. Identified datasets include the Pancreas data from the Medical Segmentation Decathlon, Pelvic Reference Data, and CT volumes with multiple organ segmentations (CT-ORG). Some of the challenges of producing an accurate model from the USC dataset include small dataset size (115 images), 2D data (as nnUNet generally performs better on 3D data), and the limited amount of public data capturing annotated CT images of bone lesions. Optimizations and improvements will be made by applying transfer learning and generative methods, including incorporating generative adversarial networks and diffusion models in order to augment the dataset. Performance with different libraries, including MONAI and custom architectures with Pytorch, will be compared. In the future, molecular correlations will be tracked with radiologic features for the purpose of multimodal composite biomarker identification. Once validated, these models will be incorporated into evaluation workflows to optimize radiologist evaluation. Our work demonstrates the challenges of applying automated image segmentation to small medical datasets and lays a foundation for techniques to improve performance. As machine learning models become increasingly incorporated into the workflow of radiologists, these findings will help improve the speed and accuracy of vertebral metastatic lesions detection.

Keywords: deep learning, image segmentation, medicine, nnUNet, prostate carcinoma, radiomics

Procedia PDF Downloads 97

Authors: Zou Yihui

Abstract:

The integration of the traditional landscape of the ancient city and the poet's emotions and symbolization into ancient poetry is the unique cultural gene and spiritual core of the historical city, and re-understanding the historical landscape pattern from the poetry is conducive to continuing the historical city context and improving the current situation of the gradual decline of the poetry of the modern historical urban landscape. Starting from Tang poetry uses semantic analysis methods、combined with text mining technology, entry mining, word frequency analysis, and cluster analysis of the landscape information of Tang Chang'an City were carried out, and the method framework for analyzing the urban landscape form based on poetry text was constructed. Nearly 160 poems describing the landscape of Tang Chang'an City were screened, and the poetic landscape characteristics of Tang Chang'an City were sorted out locally in order to combine with modern urban spatial development to continue the urban spatial context.

Keywords: Tang Chang'an City, poetic texts, semantic analysis, historical landscape

Procedia PDF Downloads 68

Authors: Budiman Budiman

Abstract:

This research aimed at describing and testing whether the students’ ability in writing analytical exposition text is increased by using comic strips at SMAN 12 Pekanbaru. The respondents of this study were the second-grade students, especially XI Science 3 academic year 2011-2012. The total number of students in this class was forty-two (42) students. The quantitative and qualitative data was collected by using writing test and observation sheets. The research finding reveals that there is a significant increase of students’ writing ability in writing analytical exposition text through comic strips. It can be proved by the average score of pre-test was 43.7 and the average score of post-test was 65.37. Besides, the students’ interest and motivation in learning are also improved. These can be seen from the increasing of students’ awareness and activeness in learning process based on observation sheets. The findings draw attention to the use of comic strips in teaching and learning is beneficial for better learning outcome.

Keywords: analytical exposition, comic strips, secondary school students, writing ability

Procedia PDF Downloads 155

Authors: Dina Zakeri, Ali Aminzad

Abstract:

Multimedia-assisted teaching results in eliminating traditional training barriers, facilitating the cognition process and upgrading learning outcomes. This study attempted to examine the effects of implementing multimedia on teaching technical translation model and on the technical text translation ability of Iranian students of translation. To fulfill the purpose of the study, a total of forty-six learners were selected out of fifty-seven participants in a higher education center in Tehran based on their scores in Preliminary English Test (PET) and were divided randomly into the experimental and control groups. Prior to the treatment, a technical text translation questionnaire was devised and then approved and validated by three assistant professors of technical fields and three assistant professors of Teaching English as a Foreign Language (TEFL) at the university. This questionnaire was administered as a pretest to both groups. Control and experimental groups were trained for five successive weeks using identical course books but with a different lesson plan that allowed employing multimedia for the experimental group only. The devised and approved questionnaire was administered as a posttest to both groups at the end of the instruction. A multivariate ANOVA was run to compare the two groups’ means on the PET, pretest and posttest. The results showed the rejection of all null hypotheses of the study and revealed that multimedia significantly improved technical text translation ability of the learners.

Keywords: multimedia, multimedia-mediated teaching, technical translation model, technical text, translation ability

Procedia PDF Downloads 129

Authors: Douglas W. Scott

Abstract:

The Allemande from J. S. Bach's Partita for solo flute (BWV 1013) presents many unique challenges for any flautist, especially in terms of segmentation analysis required to select breathing places in the first half. Without claiming to identify a 'correct' solution to this problem, this paper analyzes the section in terms of a set of techniques based around a statistical property commonly (if not ubiquitously) found in music, namely Zipf’s law. Specifically, the paper considers violations of this expected profile at various levels of analysis, an approach which has yielded interesting insights in previous studies. The investigation is then grounded by considering four actual solutions to the problem found in recordings made by different flautists, which opens up the possibility of expanding Zipfian analysis to include a consideration of inter-onset-intervals (IOIs). It is found that significant deviations from the expected Zipfian distributions can reveal and highlight stylistic choices made by different performers.

Keywords: inter-onset-interval, Partita for solo flute, BWV 1013, segmentation analysis, Zipf’s law

Procedia PDF Downloads 184

Authors: Barbara Janari

Abstract:

In this paper it is argued that the effect of the disjunctive temporality in Summertime (the third of J.M. Coetzee’s fictionalised memoirs) is two-fold: firstly, it reflects the memoir’s ambivalent, contradictory representations of place in order to emphasize the fractured sense of self growing up in South Africa during apartheid entailed for Coetzee. Secondly, it reconceives the autobiographical discourse as one that foregrounds the inherent fictionality of all texts. The memoir’s narrative is filtered through intricate textual strategies that disrupt the chronological movement of the narrative, evoking the labyrinthine ways in which the past and present intersect and interpenetrate each other. It is framed by entries from Coetzee’s Notebooks: it opens with entries that cover the years 1972–1975, and ends with a number of undated fragments from his Notebooks. Most of the entries include a short ‘memo’ at the end, added between 1999 and 2000. While the memos follow the Notebook entries in the text, they are separated by decades. Between the Notebook entries is a series of interviews conducted by Vincent, the text’s putative biographer, between 2007 and 2008, based on recollections from five people who had known Coetzee in the 1970s – a key period in John’s life as it marks both his return to South Africa after a failed emigration attempt to America, and the beginning of his writing career, with the publication of Dusklands in 1974. The relationship between the memoir’s various parts is a key feature of Coetzee’s representation of place in Summertime, which is constructed as a composite one in which the principle of reflexive referencing has to be adopted. In other words, readers have to suspend individual references temporarily until the relationships between the parts have been connected to each other. In order to apprehend meaning in the text, the disparate narrative elements have to first be tied together. In this text, then, the experience of time as ordered and chronological is ruptured. Instead, the memoir’s themes and patterns become apparent most clearly through reflexive referencing, by which relationships between disparate sections of the text are linked. The image of the fictional John that emerges from the text is a composite of this John and the author, J.M. Coetzee, and is one which embodies Coetzee’s often fraught relationship with his home country, South Africa.

Keywords: autobiography, place, reflexive referencing, temporality

Procedia PDF Downloads 76

Authors: Mian Huang, Chi Ma, Junyu Lin, William Lu

Abstract:

Introduction: Bone health attracts more attention recently and an intelligent question and answer (QA) chatbot for osteoporosis is helpful for science popularization. With Generative Pretrained Transformer (GPT) technology developing, we build an osteoporosis corpus dataset and then fine-tune LLaMA, a famous open-source GPT foundation large language model(LLM), on our self-constructed osteoporosis corpus. Evaluated by clinical orthopedic experts, our fine-tuned model outperforms vanilla LLaMA on osteoporosis QA task in Chinese. Three-dimensional quantitative computed tomography (QCT) measured bone mineral density (BMD) is considered as more accurate than DXA for BMD measurement in recent years. We develop an automatic Phantom-less QCT(PL-QCT) that is more efficient for BMD measurement since no need of an external phantom for calibration. Combined with LLM on osteoporosis, our PL-QCT provides efficient and accurate BMD measurement for our chatbot users. Material and Methods: We build an osteoporosis corpus containing about 30,000 Chinese literatures whose titles are related to osteoporosis. The whole process is done automatically, including crawling literatures in .pdf format, localizing text/figure/table region by layout segmentation algorithm and recognizing text by OCR algorithm. We train our model by continuous pre-training with Low-rank Adaptation (LoRA, rank=10) technology to adapt LLaMA-7B model to osteoporosis domain, whose basic principle is to mask the next word in the text and make the model predict that word. The loss function is defined as cross-entropy between the predicted and ground-truth word. Experiment is implemented on single NVIDIA A800 GPU for 15 days. Our automatic PL-QCT BMD measurement adopt AI-associated region-of-interest (ROI) generation algorithm for localizing vertebrae-parallel cylinder in cancellous bone. Due to no phantom for BMD calibration, we calculate ROI BMD by CT-BMD of personal muscle and fat. Results & Discussion: Clinical orthopaedic experts are invited to design 5 osteoporosis questions in Chinese, evaluating performance of vanilla LLaMA and our fine-tuned model. Our model outperforms LLaMA on over 80% of these questions, understanding ‘Expert Consensus on Osteoporosis’, ‘QCT for osteoporosis diagnosis’ and ‘Effect of age on osteoporosis’. Detailed results are shown in appendix. Future work may be done by training a larger LLM on the whole orthopaedics with more high-quality domain data, or a multi-modal GPT combining and understanding X-ray and medical text for orthopaedic computer-aided-diagnosis. However, GPT model gives unexpected outputs sometimes, such as repetitive text or seemingly normal but wrong answer (called ‘hallucination’). Even though GPT give correct answers, it cannot be considered as valid clinical diagnoses instead of clinical doctors. The PL-QCT BMD system provided by Bone’s QCT(Bone’s Technology(Shenzhen) Limited) achieves 0.1448mg/cm2(spine) and 0.0002 mg/cm2(hip) mean absolute error(MAE) and linear correlation coefficient R2=0.9970(spine) and R2=0.9991(hip)(compared to QCT-Pro(Mindways)) on 155 patients in three-center clinical trial in Guangzhou, China. Conclusion: This study builds a Chinese osteoporosis corpus and develops a fine-tuned and domain-adapted LLM as well as a PL-QCT BMD measurement system. Our fine-tuned GPT model shows better capability than LLaMA model on most testing questions on osteoporosis. Combined with our PL-QCT BMD system, we are looking forward to providing science popularization and early morning screening for potential osteoporotic patients.

Keywords: GPT, phantom-less QCT, large language model, osteoporosis

Procedia PDF Downloads 71

Authors: Nneka U. Igboeli, Maxwell O. Adibe

Abstract:

Iron and folate supplementation in pregnancy are important interventions that prevent maternal anaemia and fetal anomaly. Thus, daily oral doses of iron and folic acid are recommended throughout pregnancy as part of antenatal care. However, low adherence has been a major drawback leading to low effectiveness of these programs. The effect of mobile text message reminders to pregnant women to take their routine medications on adherence was evaluated in this study. The first 100 women who consented to the study were recruited and randomized to either receive a text message reminder on adherence to routine medications or not. Adherence was assessed using the 8-item Modified Morisky Adherence Scale (8-MMAS). The folders of successfully recruited women were tagged with the a study number assigned to each of them. The womens’ phone numbers were collected and these were used to send text messages reminders on adhering to routine drugs only to women in the intervention group. The text messages were sent three times per week for a period of four weeks with an adherence reassessment at the one month follow-up antenatal visit for recruited women. At one month follow-up, the lost to follow-up were 6 (16%) women for the intervention group and 17 (34%) for the control group. The across group mean difference in adherence score was 0.07 (-0.96 – 1.10) at baseline and 0.3 (-0.31 – 0.92) after intervention, both insignificant at p > 0.05. The within group change were increases of 0.58 (0.00 – 1.16) (p = 0.05) from baseline for the intervention group and a 0.35 (-0.51 – 1.20) (p = 0.395) for the control group. Non-significant increase in adherence scores were recorded for both groups. However, the increase in adherence scores of women in the intervention group was greater and may be potentially transformed into more positive results if the study period is increased with possibly reduced study drop-outs shows great promise for more positive results.

Keywords: adherence, mobile phone, pregnant women, reminders

Procedia PDF Downloads 175

Authors: Cheikh Bamba Dione, Alla Lo, Elhadji Mamadou Nguer, Siley O. Ba

Abstract:

In this paper, we propose two neural machine translation (NMT) systems (French-to-Wolof and Wolof-to-French) based on sequence-to-sequence with attention and transformer architectures. We trained our models on a parallel French-Wolof corpus of about 83k sentence pairs. Because of the low-resource setting, we experimented with advanced methods for handling data sparsity, including subword segmentation, back translation, and the copied corpus method. We evaluate the models using the BLEU score and find that transformer outperforms the classic seq2seq model in all settings, in addition to being less sensitive to noise. In general, the best scores are achieved when training the models on word-level-based units. For subword-level models, using back translation proves to be slightly beneficial in low-resource (WO) to high-resource (FR) language translation for the transformer (but not for the seq2seq) models. A slight improvement can also be observed when injecting copied monolingual text in the target language. Moreover, combining the copied method data with back translation leads to a substantial improvement of the translation quality.

Keywords: backtranslation, low-resource language, neural machine translation, sequence-to-sequence, transformer, Wolof

Procedia PDF Downloads 147

Authors: Mohit Modi, Rajiv Kumar, Manojraj Saxena, G. Ravi Shankar

Abstract:

Change detection of coastal ecosystem plays a vital role in monitoring and managing natural resources along the coastal regions. The present study mainly focuses on the decadal change in Kochi islands connecting the urban flatland areas and the coastal regions where sand deposits have taken place. With this, in view, the change detection has been monitored in the Kochi area to apprehend the urban growth and industrialization leading to decrease in the wetland ecosystem. The region lies between 76°11'19.134"E to 76°25'42.193"E and 9°52'35.719"N to 10°5'51.575"N in the south-western coast of India. The IRS LISS-IV satellite image has been processed using a rule-based algorithm to classify the LULC and to interpret the changes between 2005 & 2015. The approach takes two steps, i.e. extracting features as a single GIS vector layer using different parametric values and to dissolve them. The multi-resolution segmentation has been carried out on the scale ranging from 10-30. The different classes like aquaculture, agricultural land, built-up, wetlands etc. were extracted using parameters like NDVI, mean layer values, the texture-based feature with corresponding threshold values using a rule set algorithm. The objects obtained in the segmentation process were visualized to be overlaying the satellite image at a scale of 15. This layer was further segmented using the spectral difference segmentation rule between the objects. These individual class layers were dissolved in the basic segmented layer of the image and were interpreted in vector-based GIS programme to achieve higher accuracy. The result shows a rapid increase in an industrial area of 40% based on industrial area statistics of 2005. There is a decrease in wetlands area which has been converted into built-up. New roads have been constructed which are connecting the islands to urban areas as well as highways. The increase in coastal region has been visualized due to sand depositions. The outcome is well supported by quantitative assessments which will empower rich understanding of land use land cover change for appropriate policy intervention and further monitoring.

Keywords: land use land cover, multiresolution segmentation, NDVI, object based classification

Procedia PDF Downloads 187

Authors: Shreyansh Singh, Afroz Ahmed

Abstract:

Sentiment Analysis intends to naturally reveal the hidden mentality that we hold towards an entity. The total of this assumption over a populace addresses sentiment surveying and has various applications. Current text-based sentiment analysis depends on the development of word embeddings and Machine Learning models that take in conclusion from enormous text corpora. Sentiment Analysis from text is presently generally utilized for consumer loyalty appraisal and brand insight investigation. With the expansion of online media, multimodal assessment investigation is set to carry new freedoms with the appearance of integral information streams for improving and going past text-based feeling examination using the new transforms methods. Since supposition can be distinguished through compelling follows it leaves, like facial and vocal presentations, multimodal opinion investigation offers good roads for examining facial and vocal articulations notwithstanding the record or printed content. These methodologies use the Recurrent Neural Networks (RNNs) with the LSTM modes to increase their performance. In this study, we characterize feeling and the issue of multimodal assessment investigation and audit ongoing advancements in multimodal notion examination in various spaces, including spoken surveys, pictures, video websites, human-machine, and human-human connections. Difficulties and chances of this arising field are additionally examined, promoting our theory that multimodal feeling investigation holds critical undiscovered potential.

Keywords: sentiment analysis, RNN, LSTM, word embeddings

Procedia PDF Downloads 120

Authors: Sandra Rizkallah, Amir F. Atiya

Abstract:

This work presents a novel word to vector representation that is based on embedding the words into a sphere, whereby the dot product of the corresponding vectors represents the similarity between any two words. Embedding the vectors into a sphere enabled us to take into consideration the antonymity between words, not only the synonymity, because of the suitability to handle the polarity nature of words. For example, a word and its antonym can be represented as a vector and its negative. Moreover, we have managed to extract an adequate vocabulary. The obtained results show that the proposed approach can capture the essence of the language, and can be generalized to estimate a correct similarity of any new pair of words.

Keywords: natural language processing, word to vector, text similarity, text mining

Procedia PDF Downloads 276

Authors: Tae-Uk Yun, Pullip Chung, Kee-Young Kwahk, Hyunchul Ahn

Abstract:

Fake news is defined as the news articles that are intentionally and verifiably false, and could mislead readers. Spread of fake news may provoke anxiety, chaos, fear, or irrational decisions of the public. Thus, detecting fake news and preventing its spread has become very important issue in our society. However, due to the huge amount of fake news produced every day, it is almost impossible to identify it by a human. Under this context, researchers have tried to develop automated fake news detection using machine learning techniques over the past years. But, there have been no prior studies proposed an automated fake news detection method for Korean news to our best knowledge. In this study, we aim to detect Korean fake news using text mining and machine learning techniques. Our proposed method consists of two steps. In the first step, the news contents to be analyzed is convert to quantified values using various text mining techniques (topic modeling, TF-IDF, and so on). After that, in step 2, classifiers are trained using the values produced in step 1. As the classifiers, machine learning techniques such as logistic regression, backpropagation network, support vector machine, and deep neural network can be applied. To validate the effectiveness of the proposed method, we collected about 200 short Korean news from Seoul National University’s FactCheck. which provides with detailed analysis reports from 20 media outlets and links to source documents for each case. Using this dataset, we will identify which text features are important as well as which classifiers are effective in detecting Korean fake news.

Keywords: fake news detection, Korean news, machine learning, text mining

Procedia PDF Downloads 276

Authors: Kirolos Gerges Yakoub Gerges

Abstract:

Self-sustaining agricultural machines act in stochastic surroundings and therefore, should be capable of perceive the surroundings in real time. This notion can be done using image sensors blended with superior device learning, mainly Deep mastering. Deep convolutional neural networks excel in labeling and perceiving colour pix and since the fee of RGB-cameras is low, the hardware cost of accurate notion relies upon heavily on memory and computation power. This paper investigates the opportunity of designing lightweight convolutional neural networks for semantic segmentation (pixel clever class) with reduced hardware requirements, to allow for embedded usage in self-reliant agricultural machines. The usage of compression techniques, a lightweight convolutional neural community is designed to carry out actual-time semantic segmentation on an embedded platform. The community is skilled on two big datasets, ImageNet and Pascal Context, to apprehend as much as four hundred man or woman instructions. The 400 training are remapped into agricultural superclasses (e.g. human, animal, sky, road, area, shelterbelt and impediment) and the capacity to provide correct actual-time perception of agricultural environment is studied. The network is carried out to the case of self-sufficient grass mowing the usage of the NVIDIA Tegra X1 embedded platform. Feeding case-unique pics to the community consequences in a fully segmented map of the superclasses within the picture. As the network remains being designed and optimized, handiest a qualitative analysis of the technique is entire on the abstract submission deadline. intending this cut-off date, the finalized layout is quantitatively evaluated on 20 annotated grass mowing pictures. Light-weight convolutional neural networks for semantic segmentation can be implemented on an embedded platform and show aggressive performance on the subject of accuracy and speed. It’s miles viable to offer value-efficient perceptive capabilities related to semantic segmentation for autonomous agricultural machines.

Keywords: centrifuge pump, hydraulic energy, agricultural applications, irrigationaxial flux machines, axial flux applications, coreless machines, PM machinesautonomous agricultural machines, deep learning, safety, visual perception

Procedia PDF Downloads 28

Authors: Hamza Salih

Abstract:

Since translation is no longer viewed as a mere replacement of linguistic codes from one language to another, it has increasingly been considered, especially with the advent of the cultural turn in the late 70's, in relation to the broader external context in which it takes place. According to scholars in the field, the translation process is determined by the political, economic and cultural values which exert external pressures on the translator. Correspondingly, the relationship between translation as an act of re-writing the original text and ideology has already been established. This paper addresses the issue of how ideology comes into play in the translational process and what strategies the translator adopts to foreground or circumvent ideological constraints. Along with this, the paper will touch upon the notions of censorship, manipulation, subversion and domestication which are deemed of relevance to this very topic. In fact, after the domination of the empirically-oriented linguistic approaches in translation studies, the relationship between translation and ideology has to be foregrounded to draw attention to the fact that the translation process is not a mere text-to-text linguistic transfer, but, on the contrary, takes place in the midst of economic, political, cultural and religious variables, which some scholars subsume under the category ideology.

Keywords: translation, language, ideology, subversion, censorship and manipulation

Procedia PDF Downloads 249

Authors: Jiang Niu, Yue Jiang

Abstract:

With the ultimate goal to automatically generate translated texts with high quality, machine translation has made tremendous improvements. However, its translations of literary works are still plagued with problems in coherence, esp. the translation between distant language pairs. One of the causes of the problems is probably the lack of linguistic knowledge to be incorporated into the training of machine translation systems. In order to enable readers to better understand the problems of machine translation in coherence, to seek out the potential knowledge to be incorporated, and thus to improve the quality of machine translation products, this study applies Theme-Rheme structure to examine how a machine-translated literary text is organized and developed in terms of coherence. Theme-Rheme structure in Systemic Functional Linguistics is a useful tool for analysis of textual coherence. Theme is the departure point of a clause and Rheme is the rest of the clause. In a text, as Themes and Rhemes may be connected with each other in meaning, they form thematic and rhematic progressions throughout the text. Based on this structure, we can look into how a text is organized and developed in terms of coherence. Methodologically, we chose Chinese and English as the language pair to be studied. Specifically, we built a comparable corpus with two modes of English translations, viz. machine translation (MT) and human translation (HT) of one Chinese literary source text. The translated texts were annotated with Themes, Rhemes and their progressions throughout the texts. The annotated texts were analyzed from two respects, the different types of Themes functioning differently in achieving coherence, and the different types of thematic and rhematic progressions functioning differently in constructing texts. By analyzing and contrasting the two modes of translations, it is found that compared with the HT, 1) the MT features “pseudo-coherence”, with lots of ill-connected fragments of information using “and”; 2) the MT system produces a static and less interconnected text that reads like a list; these two points, in turn, lead to the less coherent organization and development of the MT than that of the HT; 3) novel to traditional and previous studies, Rhemes do contribute to textual connection and coherence though less than Themes do and thus are worthy of notice in further studies. Hence, the findings suggest that Theme-Rheme structure be applied to measuring and assessing the coherence of machine translation, to being incorporated into the training of the machine translation system, and Rheme be taken into account when studying the textual coherence of both MT and HT.

Keywords: coherence, corpus-based, literary translation, machine translation, Theme-Rheme structure

Procedia PDF Downloads 207

Authors: Sajjad Ahmed, Knut Hinkelmann, Flavio Corradini

Abstract:

Fake news detection research is still in the early stage as this is a relatively new phenomenon in the interest raised by society. Machine learning helps to solve complex problems and to build AI systems nowadays and especially in those cases where we have tacit knowledge or the knowledge that is not known. We used machine learning algorithms and for identification of fake news; we applied three classifiers; Passive Aggressive, Naïve Bayes, and Support Vector Machine. Simple classification is not completely correct in fake news detection because classification methods are not specialized for fake news. With the integration of machine learning and text-based processing, we can detect fake news and build classifiers that can classify the news data. Text classification mainly focuses on extracting various features of text and after that incorporating those features into classification. The big challenge in this area is the lack of an efficient way to differentiate between fake and non-fake due to the unavailability of corpora. We applied three different machine learning classifiers on two publicly available datasets. Experimental analysis based on the existing dataset indicates a very encouraging and improved performance.

Keywords: fake news detection, natural language processing, machine learning, classification techniques.

Procedia PDF Downloads 168

Authors: Haoqi Gao, Koichi Ogawara

Abstract:

Retinal vascular segmentation of color fundus is the basis of ophthalmic computer-aided diagnosis and large-scale disease screening systems. Early screening of fundus diseases has great value for clinical medical diagnosis. The traditional methods depend on the experience of the doctor, which is time-consuming, labor-intensive, and inefficient. Furthermore, medical images are scarce and fraught with legal concerns regarding patient privacy. In this paper, we propose a new Generative Adversarial Network based on CycleGAN for retinal fundus images. This method can generate not only synthetic fundus images but also generate corresponding segmentation masks, which has certain application value and challenge in computer vision and computer graphics. In the results, we evaluate our proposed method from both quantitative and qualitative. For generated segmented images, our method achieves dice coefficient of 0.81 and PR of 0.89 on DRIVE dataset. For generated synthetic fundus images, we use ”Toy Experiment” to verify the state-of-the-art performance of our method.

Keywords: retinal vascular segmentations, generative ad-versarial network, cyclegan, fundus images

Procedia PDF Downloads 144

Authors: Yalong Jiang, Zheru Chi

Abstract:

In this paper, we study the factors which determine the capacity of a Convolutional Neural Network (CNN) model and propose the ways to evaluate and adjust the capacity of a CNN model for best matching to a specific pattern recognition task. Firstly, a scheme is proposed to adjust the number of independent functional units within a CNN model to make it be better fitted to a task. Secondly, the number of independent functional units in the capsule network is adjusted to fit it to the training dataset. Thirdly, a method based on Bayesian GAN is proposed to enrich the variances in the current dataset to increase its complexity. Experimental results on the PASCAL VOC 2010 Person Part dataset and the MNIST dataset show that, in both conventional CNN models and capsule networks, the number of independent functional units is an important factor that determines the capacity of a network model. By adjusting the number of functional units, the capacity of a model can better match the complexity of a dataset.

Keywords: CNN, convolutional neural network, capsule network, capacity optimization, character recognition, data augmentation, semantic segmentation

Procedia PDF Downloads 155

Authors: Xiaoxiao Li, Shuangcheng Jia, Qian Li

Abstract:

Keypoint detection has always been a challenge in the field of image recognition. This paper proposes a novelty keypoint detection method which is called Multi-Scale Feature Fusion Convolutional Network with Attention (MFFCNA). We verified that the multi-scale features with the attention mechanism module have better feature expression capability. The feature fusion between different scales makes the information that the network model can express more abundant, and the network is easier to converge. On our self-made street sign corner dataset, we validate the MFFCNA model with an accuracy of 97.8% and a recall of 81%, which are 5 and 8 percentage points higher than the HRNet network, respectively. On the COCO dataset, the AP is 71.9%, and the AR is 75.3%, which are 3 points and 2 points higher than HRNet, respectively. Extensive experiments show that our method has a remarkable improvement in the keypoint recognition tasks, and the recognition effect is better than the existing methods. Moreover, our method can be applied not only to keypoint detection but also to image classification and semantic segmentation with good generality.

Keywords: keypoint detection, feature fusion, attention, semantic segmentation

Procedia PDF Downloads 120

Authors: Chun-Min Lin, Yuh-Jen Wu, Ching-Ting Chung

Abstract:

The formulation of tourism policies requires objective academic research and evidence as support, especially research from local academia. Taiwan is a small island, and its economic growth relies heavily on tourism revenue. Taiwanese government has been devoting to the promotion of the tourism industry over the past few decades. Scientific research outcomes by Taiwanese scholars may and will help lay the foundations for drafting future tourism policy by the government. In this study, a total of 120 full journal articles published between 2008 and 2016 from the Journal of Tourism and Leisure Studies (JTSL) were examined to explore the scientific research trend of tourism study in Taiwan. JTSL is one of the most important Taiwanese journals in the tourism discipline which focuses on tourism-related issues and uses traditional Chinese as the study language. The method of co-word analysis from bibliometrics approaches was employed for semantic analysis in this study. When analyzing Chinese words and phrases, word segmentation analysis is a crucial step. It must be carried out initially and precisely in order to obtain meaningful word or word chunks for further frequency calculation. A word segmentation system basing on N-gram algorithm was developed in this study to conduct semantic analysis, and 100 groups of meaningful phrases with the highest recurrent rates were located. Subsequently, co-word analysis was employed for semantic classification. The results showed that the themes of tourism research in Taiwan in recent years cover the scope of tourism education, environmental protection, hotel management, information technology, and senior tourism. The results can give insight on the related issues and serve as a reference for tourism-related policy making and follow-up research.

Keywords: bibliometrics, co-word analysis, word segmentation, tourism research, policy

Procedia PDF Downloads 229

Authors: Jaiden Xuan Schraut, Leon Liu, Yiqiao Yin

Abstract:

Computer vision in medical diagnosis has achieved a high level of success in diagnosing diseases with high accuracy. However, conventional classifiers that produce an image to-label result provides insufficient information for medical professionals to judge and raise concerns over the trust and reliability of a model with results that cannot be explained. In order to gain local insight into cancerous regions, separate tasks such as imaging segmentation need to be implemented to aid the doctors in treating patients, which doubles the training time and costs which renders the diagnosis system inefficient and difficult to be accepted by the public. To tackle this issue and drive AI-first medical solutions further, this paper proposes a multi-output network that follows a U-Net architecture for image segmentation output and features an additional convolutional neural networks (CNN) module for auxiliary classification output. Class activation maps are a method of providing insight into a convolutional neural network’s feature maps that leads to its classification but in the case of lung diseases, the region of interest is enhanced by U-net-assisted Class Activation Map (CAM) visualization. Therefore, our proposed model combines image segmentation models and classifiers to crop out only the lung region of a chest X-ray’s class activation map to provide a visualization that improves the explainability and is able to generate classification results simultaneously which builds trust for AI-led diagnosis systems. The proposed U-Net model achieves 97.61% accuracy and a dice coefficient of 0.97 on testing data from the COVID-QU-Ex Dataset which includes both diseased and healthy lungs.

Keywords: multi-output network model, U-net, class activation map, image classification, medical imaging analysis

Procedia PDF Downloads 205

Authors: Pengfei Meng, Shuangcheng Jia, Qian Li

Abstract:

We proposed a real-time high-precision semantic segmentation network based on a multi-resolution feature fusion module, the auxiliary feature extracting module, upsampling module, and atrous spatial pyramid pooling (ASPP) module. We designed a feature fusion structure, which is integrated with sufficient features of different resolutions. We also studied the effect of side-branch structure on the network and made discoveries. Based on the discoveries about the side-branch of the network structure, we used a side-branch auxiliary feature extraction layer in the network to improve the effectiveness of the network. We also designed upsampling module, which has better results than the original upsampling module. In addition, we also re-considered the locations and number of atrous spatial pyramid pooling (ASPP) modules and modified the network structure according to the experimental results to further improve the effectiveness of the network. The network presented in this paper takes the backbone network of Bisenetv2 as a basic network, based on which we constructed a network structure on which we made improvements. We named this network deep multiple-resolution bilateral networks for real-time, referred to as DMBR-Net. After experimental testing, our proposed DMBR-Net network achieved 81.2% mIoU at 119FPS on the Cityscapes validation dataset, 80.7% mIoU at 109FPS on the CamVid test dataset, 29.9% mIoU at 78FPS on the COCOStuff test dataset. Compared with all lightweight real-time semantic segmentation networks, our network achieves the highest accuracy at an appropriate speed.

Keywords: multi-resolution feature fusion, atrous convolutional, bilateral networks, pyramid pooling

Procedia PDF Downloads 154