{"title":"时间序列预测的深度学习-重点是损失函数和误差测量","authors":"Sujeeth R Malhathkar, T. S","doi":"10.1109/AIC55036.2022.9848877","DOIUrl":null,"url":null,"abstract":"Time Series Forecasting is a significant task in the modern world that finds application in several fields. The use of Deep Learning Models to perform forecasting has gained popularity in the past few years due to their superior performance in comparison to standard statistical models. 13 such models were unique in their approach to forecasting using Deep Learning models in the period 2018-2020 are studied. This work lays special focus on the use of loss function during model training and the use of error measures used to evaluate a model and compare the same with other models. We have observed that although loss functions play a role in the training process, they are not treated as rigorously as other training parameters, and several issues when it comes to reporting model performance using error metrics. The problems are highlighted and suggestions are listed.","PeriodicalId":433590,"journal":{"name":"2022 IEEE World Conference on Applied Intelligence and Computing (AIC)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"13","resultStr":"{\"title\":\"Deep Learning for Time Series Forecasting – With a focus on Loss Functions and Error Measures\",\"authors\":\"Sujeeth R Malhathkar, T. S\",\"doi\":\"10.1109/AIC55036.2022.9848877\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Time Series Forecasting is a significant task in the modern world that finds application in several fields. The use of Deep Learning Models to perform forecasting has gained popularity in the past few years due to their superior performance in comparison to standard statistical models. 13 such models were unique in their approach to forecasting using Deep Learning models in the period 2018-2020 are studied. This work lays special focus on the use of loss function during model training and the use of error measures used to evaluate a model and compare the same with other models. We have observed that although loss functions play a role in the training process, they are not treated as rigorously as other training parameters, and several issues when it comes to reporting model performance using error metrics. The problems are highlighted and suggestions are listed.\",\"PeriodicalId\":433590,\"journal\":{\"name\":\"2022 IEEE World Conference on Applied Intelligence and Computing (AIC)\",\"volume\":\"14 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-06-17\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"13\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2022 IEEE World Conference on Applied Intelligence and Computing (AIC)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/AIC55036.2022.9848877\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE World Conference on Applied Intelligence and Computing (AIC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/AIC55036.2022.9848877","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Deep Learning for Time Series Forecasting – With a focus on Loss Functions and Error Measures
Time Series Forecasting is a significant task in the modern world that finds application in several fields. The use of Deep Learning Models to perform forecasting has gained popularity in the past few years due to their superior performance in comparison to standard statistical models. 13 such models were unique in their approach to forecasting using Deep Learning models in the period 2018-2020 are studied. This work lays special focus on the use of loss function during model training and the use of error measures used to evaluate a model and compare the same with other models. We have observed that although loss functions play a role in the training process, they are not treated as rigorously as other training parameters, and several issues when it comes to reporting model performance using error metrics. The problems are highlighted and suggestions are listed.