Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773572
T. Anwar, Seemab Zakir
X-ray bone segmentation helps orthopaedic surgeons make proper decisions by separating bones from soft tissues and making the view clear. Segmenting the bones help them to analyze if the bones are in place. UNet architectures are widely used for segmentation tasks. Selecting optimal configuration help in better segmentation of bones. This paper compared different optimizers and loss functions while studying pelvic and femur bone segmentation from X-ray images. Overall, AdamW optimizers yield better performance with different loss functions than all other optimizers, including the commonly used Adam. Tversky loss shows good stable results across different optimizers in terms of the loss function. Best dice similarity coefficient and intersection over union score of 97.04 % and 96.56 % are achieved using AdamW and dice loss.
{"title":"Comparison of Loss functions and Optimizers for Multi-class X-ray Bone Segmentation","authors":"T. Anwar, Seemab Zakir","doi":"10.1109/ICAI55435.2022.9773572","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773572","url":null,"abstract":"X-ray bone segmentation helps orthopaedic surgeons make proper decisions by separating bones from soft tissues and making the view clear. Segmenting the bones help them to analyze if the bones are in place. UNet architectures are widely used for segmentation tasks. Selecting optimal configuration help in better segmentation of bones. This paper compared different optimizers and loss functions while studying pelvic and femur bone segmentation from X-ray images. Overall, AdamW optimizers yield better performance with different loss functions than all other optimizers, including the commonly used Adam. Tversky loss shows good stable results across different optimizers in terms of the loss function. Best dice similarity coefficient and intersection over union score of 97.04 % and 96.56 % are achieved using AdamW and dice loss.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123109461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773668
Saba Bashir, Kinza Rohail, Rizwan Qureshi
COVID-9 has infected nearly every country on the planet. As a result, vaccinations that can reduce our risk of contracting and spreading the COVID19 virus have been developed. As a result, each government must determine how long it will take to properly vaccinate all of its population. In this study, we built an LSTM-based prediction model to anticipate vaccination coverage in Pakistan and India. The dataset contains records of vaccine updated till January 2022. To measure the losses, we have used mean absolute error (MAE), mean absolute percentage error (MAPE), mean squared error (MSE) and Root mean squared error (RMSE). The model performs very well on training and testing datasets. This model can help government in the vaccination campaign.
{"title":"LSTM-based Model for Forecasting of COVID-19 Vaccines in Pakistan","authors":"Saba Bashir, Kinza Rohail, Rizwan Qureshi","doi":"10.1109/ICAI55435.2022.9773668","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773668","url":null,"abstract":"COVID-9 has infected nearly every country on the planet. As a result, vaccinations that can reduce our risk of contracting and spreading the COVID19 virus have been developed. As a result, each government must determine how long it will take to properly vaccinate all of its population. In this study, we built an LSTM-based prediction model to anticipate vaccination coverage in Pakistan and India. The dataset contains records of vaccine updated till January 2022. To measure the losses, we have used mean absolute error (MAE), mean absolute percentage error (MAPE), mean squared error (MSE) and Root mean squared error (RMSE). The model performs very well on training and testing datasets. This model can help government in the vaccination campaign.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114583032","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773586
Shahzad Shafiq, Luqman Ali, Wasif Khan, Rooh Ullah, Tanveer Ahmed Khan, Fady Alnaiiar
COVID-19 continues to have a devastating impact on the lives of people all over the world. Various new technologies arose in the research environment to assist mankind in surviving and living a better life. It is important to screen the infected patients in a timely and cost-effective manner to combat this disease and avoid its transmission. To achieve this aim, detection of Covid-19 from radiological evaluation of chest x-ray images using deep learning algorithms is less expensive and easily available option as it ensures fast and efficient diagnosis of the disease. Therefore, this paper presents a novel customized convolutional neural network (CNN) approach for the detection of COVID-19 from chest x-ray images. The performance of the proposed model is evaluated on three different size datasets, created from publicly available datasets. Experimental results show that the proposed model has better performance on Dataset 2. A very large increase or decrease of the number of samples in the dataset degrades the performance of the proposed model. The performance of the CNN model is compared with traditional pretrained networks namely VGG-16, VGG-19, ResNet-50 and Inception-V3. All the models show promising performance on dataset 2 which shows that optimum amount of data is enough for the model to lean features from the input data. Overall, the best validation accuracy of 97.78 was achieved by the proposed model on dataset 2.
{"title":"Covid-19 detection from X-ray images using Customized Convolutional Neural Network","authors":"Shahzad Shafiq, Luqman Ali, Wasif Khan, Rooh Ullah, Tanveer Ahmed Khan, Fady Alnaiiar","doi":"10.1109/ICAI55435.2022.9773586","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773586","url":null,"abstract":"COVID-19 continues to have a devastating impact on the lives of people all over the world. Various new technologies arose in the research environment to assist mankind in surviving and living a better life. It is important to screen the infected patients in a timely and cost-effective manner to combat this disease and avoid its transmission. To achieve this aim, detection of Covid-19 from radiological evaluation of chest x-ray images using deep learning algorithms is less expensive and easily available option as it ensures fast and efficient diagnosis of the disease. Therefore, this paper presents a novel customized convolutional neural network (CNN) approach for the detection of COVID-19 from chest x-ray images. The performance of the proposed model is evaluated on three different size datasets, created from publicly available datasets. Experimental results show that the proposed model has better performance on Dataset 2. A very large increase or decrease of the number of samples in the dataset degrades the performance of the proposed model. The performance of the CNN model is compared with traditional pretrained networks namely VGG-16, VGG-19, ResNet-50 and Inception-V3. All the models show promising performance on dataset 2 which shows that optimum amount of data is enough for the model to lean features from the input data. Overall, the best validation accuracy of 97.78 was achieved by the proposed model on dataset 2.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124093943","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773693
Abdul Subhan, Nabila Khurshid, Zarwa Shah
Wheat is at the epicenter of global food security. Extreme wheat price volatility can contribute to broader social risks in terms of food security, human development and have a significant influence on farmers' incomes in the coming years especially in developing countries like Pakistan. Wheat is not only the major staple crop of the country's food security, but it also contributes about 10.3% in agriculture which accounts for 2.2% of domestic GDP. However, the presumable intensification in climate change and macroeconomic instability is reputed as a threat to wheat price stability nationwide. Against this backdrop, this research develops a precise wheat price puzzle forecasting model using the Long- Short Term Memory Recurrent Neural Networks (LSTM-RNN) - an application of Artificial Intelligence. LSTM-RNN are proficient in handling non-linear complex systems owing to their special LSTM nodes. An assessment of the planned framework with a handful of prevailing models is also discussed. Results showed that LSTM-RNN outperformed in terms of accuracy and uncovered that wheat prices will progressively swell and shrink by 2030, which will pose menaces to the whole economy. Moreover, our proposed methodology may be used as a guiding principle for other crops as well, to fortify sustainable agriculture development by 2030.
{"title":"Uncovering Price Puzzle in the Wheat Economy of Pakistan: An Application of Artificial Neural Networks","authors":"Abdul Subhan, Nabila Khurshid, Zarwa Shah","doi":"10.1109/ICAI55435.2022.9773693","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773693","url":null,"abstract":"Wheat is at the epicenter of global food security. Extreme wheat price volatility can contribute to broader social risks in terms of food security, human development and have a significant influence on farmers' incomes in the coming years especially in developing countries like Pakistan. Wheat is not only the major staple crop of the country's food security, but it also contributes about 10.3% in agriculture which accounts for 2.2% of domestic GDP. However, the presumable intensification in climate change and macroeconomic instability is reputed as a threat to wheat price stability nationwide. Against this backdrop, this research develops a precise wheat price puzzle forecasting model using the Long- Short Term Memory Recurrent Neural Networks (LSTM-RNN) - an application of Artificial Intelligence. LSTM-RNN are proficient in handling non-linear complex systems owing to their special LSTM nodes. An assessment of the planned framework with a handful of prevailing models is also discussed. Results showed that LSTM-RNN outperformed in terms of accuracy and uncovered that wheat prices will progressively swell and shrink by 2030, which will pose menaces to the whole economy. Moreover, our proposed methodology may be used as a guiding principle for other crops as well, to fortify sustainable agriculture development by 2030.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127431396","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773725
Adnan Khan, S. Albarri, Muhammad Arslan Manzoor
Self-Supervised Learning (SSL) has enhanced the learning process of semantic representations from images. SSL has reduced the need for annotating or labelling the data by relying less on class labels during the training phase. SSL techniques dependent on Constrative Learning (CL) are acquiring prevalence because of their low dependency on training data labels. Different CL methods are producing state-of-the-art results on datasets which are used as the benchmarks for Supervised Learning. In this survey, we provide a review of CL-based methods including SimCLR, MoCo, BYOL, SwAV, SimTriplet and SimSiam. We compare these pipelines in terms of their accuracy on ImageNet and VOC07 benchmark. BYOL propose basic yet powerful architecture to accomplish 74.30 % accuracy score on image classification task. Using clustering approach SwAV outperforms other architectures by achieving 75.30 % top-1 ImageNet classification accuracy. In addition, we shed light on the importance of CL approaches which can maximise the use of huge amounts of data available today. At last, we report the impediments of current CL methodologies and emphasize the need of computationally efficient CL pipelines.
{"title":"Contrastive Self-Supervised Learning: A Survey on Different Architectures","authors":"Adnan Khan, S. Albarri, Muhammad Arslan Manzoor","doi":"10.1109/ICAI55435.2022.9773725","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773725","url":null,"abstract":"Self-Supervised Learning (SSL) has enhanced the learning process of semantic representations from images. SSL has reduced the need for annotating or labelling the data by relying less on class labels during the training phase. SSL techniques dependent on Constrative Learning (CL) are acquiring prevalence because of their low dependency on training data labels. Different CL methods are producing state-of-the-art results on datasets which are used as the benchmarks for Supervised Learning. In this survey, we provide a review of CL-based methods including SimCLR, MoCo, BYOL, SwAV, SimTriplet and SimSiam. We compare these pipelines in terms of their accuracy on ImageNet and VOC07 benchmark. BYOL propose basic yet powerful architecture to accomplish 74.30 % accuracy score on image classification task. Using clustering approach SwAV outperforms other architectures by achieving 75.30 % top-1 ImageNet classification accuracy. In addition, we shed light on the importance of CL approaches which can maximise the use of huge amounts of data available today. At last, we report the impediments of current CL methodologies and emphasize the need of computationally efficient CL pipelines.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129509153","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773469
Hamza Javaid, Aniqa Dilawari, Usman Ghani Khan, Bilal Wajid
Lying is considered a form of deception that defines one of the inevitable parts of human essence. Also, deception or lie detection has numerous applications in criminal and judicial community. Traditional practices of identifying deceit includes the monitoring of physiological signals, transcripts, visual and acoustic information with scientific techniques. In this paper, we propose a multimodal lie detection system that leverage the capabilities of novel deep learning techniques. In particular, the study investigates the importance of visual, acoustic and EEG information of a human subject for deception detection task. On the vision side, the system extracts dense optical flow features from consecutive frames in a video to monitor the facial movements. A two-stream convolution neural network utilize this visual features to detect lie or truth. Speech based deceit identification system extracts frequency distributed spectrograms from audio signals and attention augmented CNN is employed to learn changes in distribution of frequencies in speech. For lie detection with EEG signals, we utilize bidirectional long short term neural network for representation and classification of EEG data. EEG signals are represented as time series data and Bi-directional LSTM is learns the correspondences of past signals and future signals. The study performs multimodal fusion on all modalities for lie detection with best performing classifier. Experiments on Bag-Of-Lies dataset showed that the system outperformed traditional machine learning approaches with a significant difference. When all modalities are combined, the system achieves an accuracy of 83.5% in distinguishing deceptive and truthful samples.
{"title":"EEG Guided Multimodal Lie Detection with Audio-Visual Cues","authors":"Hamza Javaid, Aniqa Dilawari, Usman Ghani Khan, Bilal Wajid","doi":"10.1109/ICAI55435.2022.9773469","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773469","url":null,"abstract":"Lying is considered a form of deception that defines one of the inevitable parts of human essence. Also, deception or lie detection has numerous applications in criminal and judicial community. Traditional practices of identifying deceit includes the monitoring of physiological signals, transcripts, visual and acoustic information with scientific techniques. In this paper, we propose a multimodal lie detection system that leverage the capabilities of novel deep learning techniques. In particular, the study investigates the importance of visual, acoustic and EEG information of a human subject for deception detection task. On the vision side, the system extracts dense optical flow features from consecutive frames in a video to monitor the facial movements. A two-stream convolution neural network utilize this visual features to detect lie or truth. Speech based deceit identification system extracts frequency distributed spectrograms from audio signals and attention augmented CNN is employed to learn changes in distribution of frequencies in speech. For lie detection with EEG signals, we utilize bidirectional long short term neural network for representation and classification of EEG data. EEG signals are represented as time series data and Bi-directional LSTM is learns the correspondences of past signals and future signals. The study performs multimodal fusion on all modalities for lie detection with best performing classifier. Experiments on Bag-Of-Lies dataset showed that the system outperformed traditional machine learning approaches with a significant difference. When all modalities are combined, the system achieves an accuracy of 83.5% in distinguishing deceptive and truthful samples.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133157447","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773460
Atif Ali, Sabir Ali Changezi, H. Rizwan, Khurram Shehzad, Muhammad Usama Nazir, Muhammad Imran Naz
Geospatial information system (GIS) is a computer system used to store and manage geographic information or location-based information and input, retrieve, analyze, synthesize, and output that information. In recent years, the use of geographic information systems (GIS) in the field of modern health management has grown more and more widespread, promoting people's cognition, interpretation, prediction, and regulation of diseases in an effective manner. An infectious disease, chronic non-communicable disease, maternal and child health, environmental and food safety, disaster medical rescue, public health emergencies, and public health policy management are all covered in this article. For Medical Disaster Management, a proposed framework of GIS architecture has been developed. Meanwhile, it identifies and describes the difficulties encountered during the GIS application process, and it speculates on the potential applications of GIS in the medical field in the future.
{"title":"A GIS Architecture for Medical Disaster Management to Support Modern Healthcare Management System","authors":"Atif Ali, Sabir Ali Changezi, H. Rizwan, Khurram Shehzad, Muhammad Usama Nazir, Muhammad Imran Naz","doi":"10.1109/ICAI55435.2022.9773460","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773460","url":null,"abstract":"Geospatial information system (GIS) is a computer system used to store and manage geographic information or location-based information and input, retrieve, analyze, synthesize, and output that information. In recent years, the use of geographic information systems (GIS) in the field of modern health management has grown more and more widespread, promoting people's cognition, interpretation, prediction, and regulation of diseases in an effective manner. An infectious disease, chronic non-communicable disease, maternal and child health, environmental and food safety, disaster medical rescue, public health emergencies, and public health policy management are all covered in this article. For Medical Disaster Management, a proposed framework of GIS architecture has been developed. Meanwhile, it identifies and describes the difficulties encountered during the GIS application process, and it speculates on the potential applications of GIS in the medical field in the future.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130801195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-03-30DOI: 10.1109/ICAI55435.2022.9773772
Faisal Abdullah, Z. Jalil
In the current era of e-businesses, customer relationship management is a decisive process for selecting profitable customers and enhancing customer relationships for the betterment of the organization. However, most customer-oriented organizations face a common problem of categorizing customers, understanding the difference between them, and extracting profitable customers. In this paper, we present an approach to address these issues identifying the future and current values of customers. This helps in identifying and retaining the customers that a firm can most profitably serve. In our proposed system, we purify data through pre-processing and data cleaning, and then three key parameters i.e. recency, frequency, and monetary (RFM) are extracted from data. A Analytical Hierarchical Process is then applied to calculate the weights of RFM. These weighted RFM parameters are used for categorization of customers with the help of a fuzzy-c-mean algorithm. The validity of clusters is checked with Davies-Bouldin Index and finally, classification is done using decision tree and recommendation is given to enhance customer relationships. We evaluated the performance of our proposed system on two publicly available KDD Cup and Instacart datasets and achieved an accuracy rate of 95.5% and 94.3% respectively. The proposed system can be utilized for enhancing marketing strategies and developing new services for valuable customers.
{"title":"A Novel FCM and DT based Segmentation and Profiling Approach for Customer Relationship Management","authors":"Faisal Abdullah, Z. Jalil","doi":"10.1109/ICAI55435.2022.9773772","DOIUrl":"https://doi.org/10.1109/ICAI55435.2022.9773772","url":null,"abstract":"In the current era of e-businesses, customer relationship management is a decisive process for selecting profitable customers and enhancing customer relationships for the betterment of the organization. However, most customer-oriented organizations face a common problem of categorizing customers, understanding the difference between them, and extracting profitable customers. In this paper, we present an approach to address these issues identifying the future and current values of customers. This helps in identifying and retaining the customers that a firm can most profitably serve. In our proposed system, we purify data through pre-processing and data cleaning, and then three key parameters i.e. recency, frequency, and monetary (RFM) are extracted from data. A Analytical Hierarchical Process is then applied to calculate the weights of RFM. These weighted RFM parameters are used for categorization of customers with the help of a fuzzy-c-mean algorithm. The validity of clusters is checked with Davies-Bouldin Index and finally, classification is done using decision tree and recommendation is given to enhance customer relationships. We evaluated the performance of our proposed system on two publicly available KDD Cup and Instacart datasets and achieved an accuracy rate of 95.5% and 94.3% respectively. The proposed system can be utilized for enhancing marketing strategies and developing new services for valuable customers.","PeriodicalId":146842,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence (ICAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127535111","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：