Funding provider: This work was supported by the University of Sharjah [grant num-
ber 22020403199].
Media Release
from: University of Sharjah, United Arab Emirates
Scientists have revealed that convolutional neural networks (CNNS), a type of deep learning algorithm, demonstrate superior performance when used to detect lies and deceptions, compared to traditional non-machine learning approaches.
They demonstrate that artificial intelligence (AI) and machine learning-based methods may provide accurate predictions if one major limitation is overcome.
Researchers based at the University of Sharjah in the United Arab Emirates reported their findings in a recent review paper published in the journal Expert System with Applicationsthey present an analysis of deception detection techniques that utilize machine learning.
Their findings follow a comprehensive literature search of various databases, including Google Scholar, Elsevier, ACM Digital Library, IEE Xplore, Springer, and more, using the keyword “Deception Detection.”
The research papers selected for analysis were published between 2012 and 2023. The authors aim to extract knowledge from existing literature on how to detect and streamline deceptions, detect lies via machine learning, and compare them to traditional non-machine learning approaches.
“Our aim was to conduct a comprehensive review of publications focused on computational predictions of deceptions, using a ML (machine learning) approach, in particular. To use the papers included in the analysis, we had to adopt an ML-based approach to identify deceptions, leveraging datasets, and writing in English,” the scientist wrote.
“Through the meta-analysis, we identified a total of 98 published articles that met our criteria. The first document in the analysis dates back to 2012, with the most recent publication in 2023. In particular, about half of these papers were published after 2019.
“There are systematic review papers relating to AI-based deception detection, but we conducted a comprehensive analysis of deception detection and provided a clear overview of field contributions and limitations.”
Deception detection research has recently attracted considerable attention as scientists believe in false research, as it is possible that people rely on false research and that people relying on deceptions can lead to objective understandings of human behavior.
They note that deception and lie are used interchangeably in the literature. This categorizes deceptive behavior into different types based on its meaning, ranging from harmful to serious consequences.
An extensive review of the author shows that lies are common in everyday life, and that even those who argue that they are honest “even those who engage in occasional deceptions, the average person lies multiple times a day.” The author points out that lies can range from completely deceiving, “I didn't kill him!” The harmless white lie was used to “the outfit looked good” and to avoid embarrassing and troublesome situations.
The authors find that what they gather from lying reviews is the biggest challenge facing the legal system. “In such circumstances, mistaken truth and vice versa can have a major impact on the parties involved and society as a whole,” they point out.
They added that deception and lies research has long been a topic of interest for researchers who are concerned about their adverse effects in the clinical, ethical and legal fields. According to the authors, the text of the literature selected for this study “utilises various data collection methods, data types, and techniques, and reveals the differences in linguistic and nonverbal cues between lies and truth.”
The author's review highlights that deception studies also rely on physical signals. This is called a “physiological cues.” Scientists have used scientists to distinguish deceptive and true behavior from utterances such as pupil size, eye movement, hand position, heart rate, and things people write and say.
Scientists emphasize that there is a huge amount of research and lies in deception. Their extensive reviews cover all aspects of previous traditional methods for detecting deceptions, including the most commonly used instruments.
They add that they rely on a variety of databases, usually consisting of short statements, online articles, information disorders, interviews and supplementary indicators such as EEG (EEG) attributes, annotations, and transcription.
The authors highlight the role of recent research using AI and machine learning, whilst celebrating the traditional methods of detecting deceptions.
The science of catching liars and detecting deceptions is increasingly leaning towards AI and machine learning, which can analyze and interpret different types of data, indicating the possibility of accurately identifying falsehoods, lies, and deceptions.
This study is not limited to literature reviews. Scientists triangulate research on multiple datasets to measure the effect of gender on deception behavior. They analyze 35 short videos to determine the impact of language variation on deceptive behavior. In addition to this, we assess the impact of linguistic and cultural changes in distinguishing between truth and false conversations, in addition to a dataset covering two hours of footage.
The authors found that compared to the performance of traditional methods, a machine learning-driven approach improves the efficiency of discovering lies and discovering deceptive behaviors and statements.
However, we have identified certain restrictions that have hindered AI-based methods of detecting deceptions. According to them, one of the major limitations and gaps in the study of deception detection is that machine learning devices that cannot explain the role of culture, language, and gender cannot explain how “generalisability of findings” can limit the “generalisability of findings.”
