Fábio F. Dias | Searcharxiv

arXiv2025

Time Series Information Visualization -- A Review of Approaches and Tools

Time series data are prevalent across various domains and often encompass large datasets containing multiple time-dependent features in each sample. Exploring time-varying data is critical for data science practitioners aiming to understand dynamic behaviors and discover periodic patterns and trends. However, the analysis of such data often requires sophisticated procedures and tools. Information visualization is a communication channel that leverages human perceptual abilities to transform abstract data into visual representations. Visualization techniques have been successfully applied in the context of time series to enhance interpretability by graphically representing the temporal evolution of data. The challenge for information visualization developers lies in integrating a wide range of analytical tools into rich visualization systems that can summarize complex datasets while clearly describing the impacts of the temporal component. Such systems enable data scientists to turn raw data into understandable and potentially useful knowledge. This review examines techniques and approaches designed for handling time series data, guiding users through knowledge discovery processes based on visual analysis. We also provide readers with theoretical insights and design guidelines for considering when developing comprehensive information visualization approaches for time series, with a particular focus on time series with multiple features. As a result, we highlight the challenges and future research directions to address open questions in the visualization of time-dependent data.

cs.GR↗

arXiv2025

T-Explainer: A Model-Agnostic Explainability Framework Based on Gradients

The development of machine learning applications has increased significantly in recent years, motivated by the remarkable ability of learning-powered systems to discover and generalize intricate patterns hidden in massive datasets. Modern learning models, while powerful, often exhibit a complexity level that renders them opaque black boxes, lacking transparency and hindering our understanding of their decision-making processes. Opacity challenges the practical application of machine learning, especially in critical domains requiring informed decisions. Explainable Artificial Intelligence (XAI) addresses that challenge, unraveling the complexity of black boxes by providing explanations. Feature attribution/importance XAI stands out for its ability to delineate the significance of input features in predictions. However, most attribution methods have limitations, such as instability, when divergent explanations result from similar or the same instance. This work introduces T-Explainer, a novel additive attribution explainer based on the Taylor expansion that offers desirable properties such as local accuracy and consistency. We demonstrate T-Explainer's effectiveness and stability over multiple runs in quantitative benchmark experiments against well-known attribution methods. Additionally, we provide several tools to evaluate and visualize explanations, turning T-Explainer into a comprehensive XAI framework.

cs.LG↗