From Pixels to Predictions: Spectrogram and Vision Transformer for Better Time Series Forecasting

Abstract

Time series forecasting plays a crucial role in decision-making across various domains, but it presents significant challenges. Recent studies have explored image-driven approaches using computer vision models to address these challenges, often employing lineplots as the visual representation of time series data. In this paper, we propose a novel approach that uses time-frequency spectrograms as the visual representation of time series data. We introduce the use of a vision transformer for multimodal learning, showcasing the advantages of our approach across diverse datasets from different domains. To evaluate its effectiveness, we compare our method against statistical baselines (EMA and ARIMA), a state-of-the-art deep learning-based approach (DeepAR), other visual representations of time series data (lineplot images), and an ablation study on using only the time series as input. Our experiments demonstrate the benefits of utilizing spectrograms as a visual representation for time series data, along with the advantages of employing a vision transformer for simultaneous learning in both the time and frequency domains.

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

“…In the financial domain, Wang et al [17] harnessed the Transformer for predicting stock market indices, outperforming other conventional deep learning models. Zeng et al [18] combined CNN with Transformer to establish a time series model (CTTS) capturing both short-term patterns and long-term dependencies. Xu et al [19] introduced a novel Transformer model for financial time series prediction, simplifying the Transformer and integrating the attention mechanism.…”

Machine Learning Methods in Algorithmic Trading: An Experimental Evaluation of Supervised Learning Techniques for Stock Price

“…Recent advancements in Machine Learning (ML) have illuminated the long-standing quest for precise price predictions. The emergence of transformative technologies has injected fresh energy into forecasting [2]. Among the notable technologies are Transformers [3], Long Short-Term Memory (LSTM) [4], Simple Recurrent Neural Networks (Simple RNN) [5], NHits [6], and NBeats [7].…”