Social media platforms, such as Twitter, have become powerful sources of information on people's perception of major events. Many people use Twitter to express their views on various issues and events and use it to develop their opinion on the diverse economic, political, technical, and social occurrences related to their daily lives. Spam and non-relevant tweets are a major challenge for Twitter trend detection. Saudi Arabia is a top ranked country in Twitter usage worldwide, and in recent years has experienced difficulties due to the use and rise of hashtags based on misleading tweets and spam. The goal of this paper is to apply machine learning techniques to identify spam on the Saudi tweets collected to the end of 2020. To date, spam detection on Twitter data has been mostly done in English, leaving other major languages, such as Arabic, insufficiently covered. Additionally, publicly accessible Arabic Twitter datasets are hard to find. For our research, we use eight Twitter datasets on some significant topics in politics, health, national affairs, economy, and sport, to train and evaluate different machine learning algorithms, with a focus on two feature generation techniques based on N-grams and Word2Vec embeddings. One contribution of this paper is providing these new labelled datasets with embeddings. The experimental results show improvement from using embeddings over N-grams in more balanced datasets vs. more unbalanced ones. We also find a superior performance of the Random Forest algorithm over other algorithms in most experiments.
{"title":"N-gram and Word2Vec Feature Engineering Approaches for Spam Recognition on Some Influential Twitter Topics in Saudi Arabia","authors":"Ahmed M. Balfagih, Vlado Keselj, Stacey Taylor","doi":"10.1145/3546157.3546173","DOIUrl":"https://doi.org/10.1145/3546157.3546173","url":null,"abstract":"Social media platforms, such as Twitter, have become powerful sources of information on people's perception of major events. Many people use Twitter to express their views on various issues and events and use it to develop their opinion on the diverse economic, political, technical, and social occurrences related to their daily lives. Spam and non-relevant tweets are a major challenge for Twitter trend detection. Saudi Arabia is a top ranked country in Twitter usage worldwide, and in recent years has experienced difficulties due to the use and rise of hashtags based on misleading tweets and spam. The goal of this paper is to apply machine learning techniques to identify spam on the Saudi tweets collected to the end of 2020. To date, spam detection on Twitter data has been mostly done in English, leaving other major languages, such as Arabic, insufficiently covered. Additionally, publicly accessible Arabic Twitter datasets are hard to find. For our research, we use eight Twitter datasets on some significant topics in politics, health, national affairs, economy, and sport, to train and evaluate different machine learning algorithms, with a focus on two feature generation techniques based on N-grams and Word2Vec embeddings. One contribution of this paper is providing these new labelled datasets with embeddings. The experimental results show improvement from using embeddings over N-grams in more balanced datasets vs. more unbalanced ones. We also find a superior performance of the Random Forest algorithm over other algorithms in most experiments.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130856055","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}
Attributed graphs are used to model real-life systems in many domains such as social science, biology, etc. Link prediction is an important task on attributed graph with a wide range of useful applications. Simple link prediction approaches have limitation in their capability to capture network topology and node attributes. Graph Neural Networks (GNNs) provide an efficient framework incorporating node attributes and connectivity to produce informative embeddings for many downstream tasks including link prediction. In this work, we study two variants of GNNs, namely Simple Graph Convolution (SGC) and its extension for link prediction on three citation datasets. While it is fast and efficient, our model is insufficient to capture the complex node connectivities. On the other hand, imposing regularization reduces overfitting and improves model performance.
{"title":"Link prediction with Simple Graph Convolution and regularized Simple Graph Convolution","authors":"Patrick Pho, Alexander V. Mantzaris","doi":"10.1145/3546157.3546163","DOIUrl":"https://doi.org/10.1145/3546157.3546163","url":null,"abstract":"Attributed graphs are used to model real-life systems in many domains such as social science, biology, etc. Link prediction is an important task on attributed graph with a wide range of useful applications. Simple link prediction approaches have limitation in their capability to capture network topology and node attributes. Graph Neural Networks (GNNs) provide an efficient framework incorporating node attributes and connectivity to produce informative embeddings for many downstream tasks including link prediction. In this work, we study two variants of GNNs, namely Simple Graph Convolution (SGC) and its extension for link prediction on three citation datasets. While it is fast and efficient, our model is insufficient to capture the complex node connectivities. On the other hand, imposing regularization reduces overfitting and improves model performance.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131029290","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}
After the introduction of blockchain as a cryptocurrency platform, researchers and industry leaders have come up with novel ways of utilizing the technology. One emerging use case for blockchain is access control, since it solves the problem of trust deficit while being distributed, auditable and private. This paper lists various access control models that have been proposed or implemented on blockchain platforms. It also provides an analysis of the performance of some of these models. Analysis shows that access control models are progressing from the traditional identity-based systems to role/attribute-based systems with observed structural shifts from centralized to decentralized. Key benefits of blockchain-based access control systems noted are improved transparency over access control and logging; improved, more complex policy management; and the possibility to implement access controls in trust-less systems.
{"title":"Access Control using Blockchain: A Taxonomy and Review","authors":"S. Malik, M. A. Shah","doi":"10.1145/3546157.3546165","DOIUrl":"https://doi.org/10.1145/3546157.3546165","url":null,"abstract":"After the introduction of blockchain as a cryptocurrency platform, researchers and industry leaders have come up with novel ways of utilizing the technology. One emerging use case for blockchain is access control, since it solves the problem of trust deficit while being distributed, auditable and private. This paper lists various access control models that have been proposed or implemented on blockchain platforms. It also provides an analysis of the performance of some of these models. Analysis shows that access control models are progressing from the traditional identity-based systems to role/attribute-based systems with observed structural shifts from centralized to decentralized. Key benefits of blockchain-based access control systems noted are improved transparency over access control and logging; improved, more complex policy management; and the possibility to implement access controls in trust-less systems.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"39 1-2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123662883","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}
Traffic sign recognition is an integral part of future autonomous driving systems. Deep learning has been applied in this task, while the performance of the recent vision Transformers is unexplored. In this study, eight different vision Transformers are validated in three real-world traffic sign datasets for the first time. The experimental results demonstrate that the best vision Transformer has a performance between the pre-trained DenseNet and the DenseNet trained from scratch. Besides, the best vision Transformers model has less training time compared to DenseNet.
{"title":"Traffic Sign Recognition with Vision Transformers","authors":"Haolan Wang","doi":"10.1145/3546157.3546166","DOIUrl":"https://doi.org/10.1145/3546157.3546166","url":null,"abstract":"Traffic sign recognition is an integral part of future autonomous driving systems. Deep learning has been applied in this task, while the performance of the recent vision Transformers is unexplored. In this study, eight different vision Transformers are validated in three real-world traffic sign datasets for the first time. The experimental results demonstrate that the best vision Transformer has a performance between the pre-trained DenseNet and the DenseNet trained from scratch. Besides, the best vision Transformers model has less training time compared to DenseNet.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122482040","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}
Uzair Iqbal, Teh Ying Wah, Muhammad Habib Ur Rehman, Muhammad Bilal, Adeel Ahmed
In the cardiovascular diseases, early detection and identification of the relationship between different diseases are still open problems for cardiologists. In this paper, we propose a novel scheme for behavioral detection in electrocardiography data named as Behavioral detection-Electrography. The Behavioral detection-Electrography is used for early detection of abnormalities in electrography especially myocardial infarction. The Behavioral detection-Electrography embeds the two-tier architecture in which we integrate the behavioral relationship concepts with myocardial detection algorithm. In future the highlighted integral scheme will help us to identify the nature of cardiovascular diseases either it's normal or abnormal
{"title":"BD-ECG: Identification of Myocardial Infarction in ECG via Behavior Coupling","authors":"Uzair Iqbal, Teh Ying Wah, Muhammad Habib Ur Rehman, Muhammad Bilal, Adeel Ahmed","doi":"10.1145/3546157.3546169","DOIUrl":"https://doi.org/10.1145/3546157.3546169","url":null,"abstract":"In the cardiovascular diseases, early detection and identification of the relationship between different diseases are still open problems for cardiologists. In this paper, we propose a novel scheme for behavioral detection in electrocardiography data named as Behavioral detection-Electrography. The Behavioral detection-Electrography is used for early detection of abnormalities in electrography especially myocardial infarction. The Behavioral detection-Electrography embeds the two-tier architecture in which we integrate the behavioral relationship concepts with myocardial detection algorithm. In future the highlighted integral scheme will help us to identify the nature of cardiovascular diseases either it's normal or abnormal","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123542477","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}
Deep learning methods have been developed for handwritten digit classification. However, these methods work as ‘black-boxes’ and need large training data. In this study, an explainable feature extraction method is developed for handwritten digit classification. The features of the digit image include horizontal, vertical, and orthogonal lines as well as full or semi-circles. In our proposed method, such features are extracted using 10 neurons as computational units. Specifically, the neurons store the features through network training and store them inside the neurons in a non-synaptic memory manner. Following that, the trained neurons are used for the retrieval of information from test images to assign them to digit categories. Our method shows an accuracy of 75 % accuracy using 0.016 % of the training data and achieves a high accuracy of 86 % using one epoch of whole training data of the MNIST dataset. To the best of our knowledge, this is the first model that stores information inside a few single neurons (i.e., non-synaptic memory) instead of storing the information in synapses of connected feed-forward layers. Due to enabling single neurons to compute individually, it is expected that such a class of neural networks can be combined with synaptic memory architectures that we expect to show higher performance compared to traditional neural networks.
{"title":"A Nonsynaptic Memory Based Neural Network for Hand-Written Digit Classification Using an Explainable Feature Extraction Method","authors":"F. Faghihi, Siqi Cai, A. Moustafa, Hany Alashwal","doi":"10.1145/3546157.3546168","DOIUrl":"https://doi.org/10.1145/3546157.3546168","url":null,"abstract":"Deep learning methods have been developed for handwritten digit classification. However, these methods work as ‘black-boxes’ and need large training data. In this study, an explainable feature extraction method is developed for handwritten digit classification. The features of the digit image include horizontal, vertical, and orthogonal lines as well as full or semi-circles. In our proposed method, such features are extracted using 10 neurons as computational units. Specifically, the neurons store the features through network training and store them inside the neurons in a non-synaptic memory manner. Following that, the trained neurons are used for the retrieval of information from test images to assign them to digit categories. Our method shows an accuracy of 75 % accuracy using 0.016 % of the training data and achieves a high accuracy of 86 % using one epoch of whole training data of the MNIST dataset. To the best of our knowledge, this is the first model that stores information inside a few single neurons (i.e., non-synaptic memory) instead of storing the information in synapses of connected feed-forward layers. Due to enabling single neurons to compute individually, it is expected that such a class of neural networks can be combined with synaptic memory architectures that we expect to show higher performance compared to traditional neural networks.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114665324","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}
Based on the job demands-resources model, this paper explores the mediating role of psychological empowerment in the process of transformational leadership impacts the effects of employees’ emotional labor. We report empirical results indicating that transformational leadership has a strong negative relationship to surface behavior and a strong positive relationship to deep behavior, while psychological empowerment mediates the relationship between transformational leadership and employees' emotional labor. Specifically, transformational leadership can indirectly influence the deep acting of front-line employees through meaning, influence, self-determination and self-efficacy; Transformational leadership can indirectly influence front-line employees' surface acting through influence, self-determination and self-efficacy, while meaning has no significant influence on surface acting.
{"title":"Influence of Transformational Leadership on Emotional Labor of Employees: —Mediating Role of Psychological Empowerment","authors":"Pengfei Cheng, Jingxuan Jiang, Shasha Tian","doi":"10.1145/3546157.3546179","DOIUrl":"https://doi.org/10.1145/3546157.3546179","url":null,"abstract":"Based on the job demands-resources model, this paper explores the mediating role of psychological empowerment in the process of transformational leadership impacts the effects of employees’ emotional labor. We report empirical results indicating that transformational leadership has a strong negative relationship to surface behavior and a strong positive relationship to deep behavior, while psychological empowerment mediates the relationship between transformational leadership and employees' emotional labor. Specifically, transformational leadership can indirectly influence the deep acting of front-line employees through meaning, influence, self-determination and self-efficacy; Transformational leadership can indirectly influence front-line employees' surface acting through influence, self-determination and self-efficacy, while meaning has no significant influence on surface acting.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128544357","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}
Crowd control is a public policy technique in which massive crowds are handled in order to avoid the emergence of possible issues or threats caused by COVID-19 and over-crowding. In this pandemic, social distancing is critical as there is a high chance of being infected in a crowd. With mounting fears about public disease transmission, the significance of crowd monitoring is crucial in these testing times. In the existing system, the model takes more time and resources to process the data from the crowd control application thus resulting in delayed prediction. Early prediction of the crowd level will help people and other government agencies to control and monitor the crowd. Hence, the main goal of the proposed system is to process a large amount of input from the crowd control application in minimal time using Dynamic Task Scheduling (Dask) based Hadoop framework in a multi-node docker cluster. The multi-node cluster processes the input data in different clusters. Each cluster data is fed to model for prediction and forecasting the count of crowd at a location. The models considered for evaluation are RNN_LSTM and ARIMA. The results shown that RNN_LSTM model has provided better accuracy of 97% compared to the ARIMA of 89%. The results show that the prediction performance of RNN_LSTM has shown 40% decrease in Mean Absolute Error (MAE) and 30% decrease in Root Mean Squared Error (RMSE) over the existing ARIMA model. The proposed system is available as an application to the public and enable them to decide whether to visit a particular place or not.
{"title":"Docker Container based Crowd Control Analysis Using Dask Hadoop Framework","authors":"G. RadhikaE., Jai Bhaarath, Naveen, Ritesh Nirmal","doi":"10.1145/3546157.3546159","DOIUrl":"https://doi.org/10.1145/3546157.3546159","url":null,"abstract":"Crowd control is a public policy technique in which massive crowds are handled in order to avoid the emergence of possible issues or threats caused by COVID-19 and over-crowding. In this pandemic, social distancing is critical as there is a high chance of being infected in a crowd. With mounting fears about public disease transmission, the significance of crowd monitoring is crucial in these testing times. In the existing system, the model takes more time and resources to process the data from the crowd control application thus resulting in delayed prediction. Early prediction of the crowd level will help people and other government agencies to control and monitor the crowd. Hence, the main goal of the proposed system is to process a large amount of input from the crowd control application in minimal time using Dynamic Task Scheduling (Dask) based Hadoop framework in a multi-node docker cluster. The multi-node cluster processes the input data in different clusters. Each cluster data is fed to model for prediction and forecasting the count of crowd at a location. The models considered for evaluation are RNN_LSTM and ARIMA. The results shown that RNN_LSTM model has provided better accuracy of 97% compared to the ARIMA of 89%. The results show that the prediction performance of RNN_LSTM has shown 40% decrease in Mean Absolute Error (MAE) and 30% decrease in Root Mean Squared Error (RMSE) over the existing ARIMA model. The proposed system is available as an application to the public and enable them to decide whether to visit a particular place or not.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115148439","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}
With the spread of Covid-19, there has been a shift towards distance learning and teachers find it difficult to keep track of students who are attentive during class. Unlike before, where the traditional classroom environment helped teachers keep track of the students who are not fully concentrating during the lessons. This shift to online learning has made teachers find it much more difficult to keep track of students who are idling during their lecture period. For this the following solution is proposed to introduce an extension to help teachers integrate with existing video conferencing platforms. This solution will help teachers to know whether the student has been attentive during class, by keeping track of their peripheral device movements, such as mouse movements or keystrokes. Previous studies have been conducted to keep track of student's eye movement and browser history, but no solution has been developed to easily ‘plug and play’ into an existing platform for teachers to get real time progress of a student's interaction to the lecture. The main objective of this research will be to help enhance the learning experience of a student by keeping the teacher aware of the student's progress just like in a traditional classroom environment.
{"title":"Effective online learning management system to improve and enhance the online learning and student engagement experience.: This document contains how the proposed solution, an extension, can help enhance distance learning.","authors":"T. Ginige, Shenal Thilaksha Vanderwall","doi":"10.1145/3546157.3546172","DOIUrl":"https://doi.org/10.1145/3546157.3546172","url":null,"abstract":"With the spread of Covid-19, there has been a shift towards distance learning and teachers find it difficult to keep track of students who are attentive during class. Unlike before, where the traditional classroom environment helped teachers keep track of the students who are not fully concentrating during the lessons. This shift to online learning has made teachers find it much more difficult to keep track of students who are idling during their lecture period. For this the following solution is proposed to introduce an extension to help teachers integrate with existing video conferencing platforms. This solution will help teachers to know whether the student has been attentive during class, by keeping track of their peripheral device movements, such as mouse movements or keystrokes. Previous studies have been conducted to keep track of student's eye movement and browser history, but no solution has been developed to easily ‘plug and play’ into an existing platform for teachers to get real time progress of a student's interaction to the lecture. The main objective of this research will be to help enhance the learning experience of a student by keeping the teacher aware of the student's progress just like in a traditional classroom environment.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122248806","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}
With time-series data being prevalent everywhere, there is a need to predict this data accurately. This kind of data includes weather data, financial data such as stock price, and cryptocurrency price. Most of the trades in the stock market in this day and age are being made using artificial intelligence. An estimated 50% of trades were done using an algorithm, which increased to 60% in 2020 [1]. This highlights the demand for reliable and accurate predictions. The prediction of the price is very challenging. Some success has been seen when predicting stock prices, but not many studies have been done on cryptocurrency. Cryptocurrency, specifically Bitcoin, has seen a substantial increase in popularity, and the price has reflected this popularity. The price also follows patterns specifically when reaching new all-time highs. In this work, an Artificial intelligence is created and trained on the previous data to observe these patterns and predict the next price. The artificial intelligence chosen for this subject is Long short-term memory (LSTM). LSTMs are capable of finding patterns in time series data. LSTM solves the vanishing gradient problem present in the RNN (Recurrent Neural Network). The Market Price of Bitcoin is used as input here. The data values for input range from 20,000 up to 65,000 in testing. Once an optimal starting point is found, there is an 80/20 split of data, 80 percent of the data is used for training and 20 is used for testing. With the data being split, one of the most important jobs is figuring out the optimal lags (how far back into the past) when used to predict values. This range for this experiment is set to ten previous price days. Epochs (number of iterations) and Batch size (how much of the training data is used per epoch) are tested at different values to find optimal solutions. With batch size values such that batchSize ∈ {20, 21…26} and epochs such that epochs ∈ {10, 20….70}. Overfitting is hard to detect and thus can be an issue with too many epochs and smaller batch sizes (smaller means more of the training data is used). Too little and the LSTM will not learn the data patterns and thus will not have good accuracy. This is why different configurations are used in the experiment to maximize accuracy. This LSTM was used to achieve a Mean Absolute Percentage Error score of 3.23% and a Root Mean Squared Error score of 1892.87 when predicting next-day prices throughout 350.
{"title":"Long Short-Term Memory for Bitcoin Price Prediction","authors":"Jordan Jones, Doga Demirel","doi":"10.1145/3546157.3546162","DOIUrl":"https://doi.org/10.1145/3546157.3546162","url":null,"abstract":"With time-series data being prevalent everywhere, there is a need to predict this data accurately. This kind of data includes weather data, financial data such as stock price, and cryptocurrency price. Most of the trades in the stock market in this day and age are being made using artificial intelligence. An estimated 50% of trades were done using an algorithm, which increased to 60% in 2020 [1]. This highlights the demand for reliable and accurate predictions. The prediction of the price is very challenging. Some success has been seen when predicting stock prices, but not many studies have been done on cryptocurrency. Cryptocurrency, specifically Bitcoin, has seen a substantial increase in popularity, and the price has reflected this popularity. The price also follows patterns specifically when reaching new all-time highs. In this work, an Artificial intelligence is created and trained on the previous data to observe these patterns and predict the next price. The artificial intelligence chosen for this subject is Long short-term memory (LSTM). LSTMs are capable of finding patterns in time series data. LSTM solves the vanishing gradient problem present in the RNN (Recurrent Neural Network). The Market Price of Bitcoin is used as input here. The data values for input range from 20,000 up to 65,000 in testing. Once an optimal starting point is found, there is an 80/20 split of data, 80 percent of the data is used for training and 20 is used for testing. With the data being split, one of the most important jobs is figuring out the optimal lags (how far back into the past) when used to predict values. This range for this experiment is set to ten previous price days. Epochs (number of iterations) and Batch size (how much of the training data is used per epoch) are tested at different values to find optimal solutions. With batch size values such that batchSize ∈ {20, 21…26} and epochs such that epochs ∈ {10, 20….70}. Overfitting is hard to detect and thus can be an issue with too many epochs and smaller batch sizes (smaller means more of the training data is used). Too little and the LSTM will not learn the data patterns and thus will not have good accuracy. This is why different configurations are used in the experiment to maximize accuracy. This LSTM was used to achieve a Mean Absolute Percentage Error score of 3.23% and a Root Mean Squared Error score of 1892.87 when predicting next-day prices throughout 350.","PeriodicalId":422215,"journal":{"name":"Proceedings of the 6th International Conference on Information System and Data Mining","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128288596","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：