Time-Series Neural Network: A High-Accuracy Time-Series Forecasting Method Based on Kernel Filter and Time Attention

Abstract

This research introduces a novel high-accuracy time-series forecasting method, namely the Time Neural Network (TNN), which is based on a kernel filter and time attention mechanism. Taking into account the complex characteristics of time-series data, such as non-linearity, high dimensionality, and long-term dependence, the TNN model is designed and implemented. The key innovations of the TNN model lie in the incorporation of the time attention mechanism and kernel filter, allowing the model to allocate different weights to features at each time point, and extract high-level features from the time-series data, thereby improving the model’s predictive accuracy. Additionally, an adaptive weight generator is integrated into the model, enabling the model to automatically adjust weights based on input features. Mainstream time-series forecasting models such as Recurrent Neural Networks (RNNs) and Long Short-Term Memory Networks (LSTM) are employed as baseline models and comprehensive comparative experiments are conducted. The results indicate that the TNN model significantly outperforms the baseline models in both long-term and short-term prediction tasks. Specifically, the RMSE, MAE, and R2 reach 0.05, 0.23, and 0.95, respectively. Remarkably, even for complex time-series data that contain a large amount of noise, the TNN model still maintains a high prediction accuracy.

Machine Learning-Based Simulation of the Air Conditioner Operating Time in Concrete Structures with Bayesian Thresholding

“…In addition, the test data also showed close fitting. With simple structures of data, this can readily occur in RNN studies [47,48]. The close fitting and following exactly along with the raw data indicates why RNN is widely using in time series forecasting [47,48].…”

Glu-Ensemble: An ensemble deep learning framework for blood glucose forecasting in type 2 diabetes patients

“…3 ). The EGA grid zones are structured according to the Parkes Error Grid (PEG), a commonly utilized framework for evaluating the clinical implications of blood glucose readings in T2DM patients [ 48 , 49 ]. The PEG segments the blood glucose range into five zones (A–E), where each zone corresponds to varying degrees of clinical risk associated with the glucose readings.…”

Advancing Financial Forecasts: A Deep Dive into Memory Attention and Long-Distance Loss in Stock Price Predictions

“…Support Vector Machine (SVM) [25]: the Support Vector Machine is a powerful model widely used for classification and regression problems. Time-series Neural Network [26]: a Transformer-based time series analysis model that employs the self-attention mechanism to capture long-term dependencies. Temporal Transformer [27]: designed specifically for time series data, this Transformer model integrates traditional time series analysis techniques.…”

“…Recall measures the proportion of actual positive instances that were correctly predicted as positive. Its formula is: Recall = True Positives (TP) True Positives (TP) + False Negatives (FN) (26) In this case, False Negatives (FN) are the instances that were incorrectly predicted as negative.…”