Experiments with truth using Machine Learning: Spectral analysis and explainable classification of synthetic, false, and genuine information

TL;DR

This paper investigates the spectral and explainability aspects of classifying synthetic, false, and genuine information, revealing the close intertwining of misinformation with genuine data and the limitations of current ML methods.

Contribution

It provides a comprehensive spectral and explainability analysis of misinformation detection, highlighting the challenges and limitations of existing machine learning approaches.

Findings

ML algorithms struggle to effectively distinguish misinformation from genuine information

Spectral analysis reveals close similarities between false and genuine data

Explainability methods show intertwined features in misinformation and genuine content

Abstract

Misinformation is still a major societal problem and the arrival of Large Language Models (LLMs) only added to it. This paper analyzes synthetic, false, and genuine information in the form of text from spectral analysis, visualization, and explainability perspectives to find the answer to why the problem is still unsolved despite multiple years of research and a plethora of solutions in the literature. Various embedding techniques on multiple datasets are used to represent information for the purpose. The diverse spectral and non-spectral methods used on these embeddings include t-distributed Stochastic Neighbor Embedding (t-SNE), Principal Component Analysis (PCA), and Variational Autoencoders (VAEs). Classification is done using multiple machine learning algorithms. Local Interpretable Model-Agnostic Explanations (LIME), SHapley Additive exPlanations (SHAP), and Integrated Gradients are used for the explanation of the classification. The analysis and the explanations generated show that misinformation is quite closely intertwined with genuine information and the machine learning algorithms are not as effective in separating the two despite the claims in the literature.