数据是electricity这个数据集,比较常用的写time series forecasting 的 paper常用的测试数据,数据量比较大,相对来说得到的结果更可靠一些。评估指标使用mae和rmse,loss使用mae。 考虑到不同人使用nn的姿势不太一样可能导致同一个模型不同设定背景下差异较大,所以基本上是从kaggle的top solution中找到各种个人觉得比较magic...
patch in transformer-based models。局部信息很重要时,patch是不可缺少的一部分,在Bert中将每个单词作为一个token(patch),而在视觉transformer中,patch则是输入图片分割成均匀块后的对象。 transformer-based long-term time series forecasting。模型大多侧重于设计新的机制来降低原有注意机制的复杂性,从而达到更好的...
1.Transformer-based Model 论文题目:Deep Transformer Models for Time Series Forecasting: The Influenza Prevalence Case 下载地址:https://arxiv.org/pdf/2001.08317.pdf 在我印象里,这篇文章应该是第一篇正式发表的在时间序列预测中使用Transformer结构的文章。其整体结构就是直接将Transformer的En...
# step 4: add temporal features based on freq of the dataset # these serve as positional encodings AddTimeFeatures( start_field=FieldName.START, target_field=FieldName.TARGET, output_field=FieldName.FEAT_TIME, time_features=time_features_from_frequency_str(freq), pred_length=config.prediction_le...
此数据集是 Monash Time Series Forecasting 存储库的一部分,该存储库收纳了是来自多个领域的时间序列数据集。它可以看作是时间序列预测的 GLUE 基准。 from datasets import load_datasetdataset = load_dataset("monash_tsf", "tourism_monthly") 可以看出,数据集包...
Attention-based models for speech recognition 2.BERT BERT: pre-training of deep bidirectional transformers for language understanding 3.AST Adversarial sparse transformer for time series forecasting 4.Informer Informer: beyond efficient transformer for long sequence time-series forecasting ...
The iTransformer is applied on the inverted shape of the series. That way, the whole feature across time is tokenized and embedded. Image by Y. Liu, T. Hu, H. Zhang, H. Wu, S. Wang, L. Ma, M. Long fromiTransformer: Inverted Transformers Are Effective for Time Series Forecasting. ...
论文题目: Informer: Beyond efficient transformer for long sequence timeseries forecasting(AAAI 2021) 下载地址:https://arxiv.org/pdf/2012.07436.pdf Informer针对长周期时间序列预测,主要从效率角度对Transformer进行了优化,这篇工作也是AAAI 2021的best paper。对于长周期时间序列,输入的维度很高,而Transformer的时间...
本文中了 2023 ICLR。作为 Transformer-based 预测模型,它是和计算机视觉中的 ViT 最相似的一篇论文(文章标题也很像)。它成功超过了 DLinear,也证明了 DLinear 中 Transformer可能不适合于序列预测任务的声明是值得商榷的。 论文标题: A Time Series is Worth 64 Words: Long-term Forecasting with Transformers ...
Attention-based models for speech recognition 2.BERT BERT: pre-training of deep bidirectional transformers for language understanding 3.AST Adversarial sparse transformer for time series forecasting 4.Informer Informer: beyond efficient transformer for long sequence time-...