An Efficient Approach to Regression Problems with Tensor Neural Networks

As a widely employed statistical method across various domains, regression analysis predicts or models the relationship between independent and dependent variables [1]. To accommodate data of diverse scales and characteristics, numerous regression methods have been developed, resulting in favorable practical outcomes [2,3]. Despite their success, ongoing efforts aim to devise more efficient algorithms to enhance both accuracy and interpretability. Technological advancements in various industries have led to increasingly complex, high-dimensional, and structured datasets. These datasets often contain information from diverse domains such as spatial, imagery, and spectral data. Such data should be analyzed as a unified and structured entity rather than as a mere collection of data points.