EAGER: SaTC-EDU: Improving Cybersecurity Education for Adolescents with Autism Through Automated Augmented Self-Monitoring Applications

Social media and online gaming sites are ubiquitous platforms that provide peer interaction and social development for adolescents but also pose potential threats to health and safety. For example, cyberbullying has been connected to negative mental health outcomes for adolescents. Addressing cyberbullying and cybersecurity issues in these spaces is therefore essential for healthy social development.

While cyberbullying is of concern for all adolescents, adolescents with Autism Spectrum Disorder (ASD) are especially vulnerable and are at a higher risk of becoming victims of bullying and theft of personal data. This research addresses these issues by creating and employing personalized virtual companions, whose interactions with the adolescent are guided by artificial intelligence (AI).

The personalized virtual companion will guide adolescents with ASD to improve their self-protection strategies and increase their cybertechnology knowledge. The goal of the project is to help prepare these individuals to become productive and satisfied members of a modern, inclusive, and technology-savvy workforce.

This project will address the problem of cyberbullying and cybersecurity for adolescents with ASD by developing tools based on deep learning algorithms that employ multimodal data, e.g., facial expression, body posture, breathing rate, skin temperature, heart rate, heart rate variability, eye gaze, verbalization, and vocalization. Fusing data from these diverse sources is expected to enable creation of a virtual learning environment that is inhabited by autonomous agents, some of which have bullying or privacy-invading tendencies, and one of which is a personalized companion that seeks to help its human friend navigate the Internet.

The AI companion must adapt its interactions based on understanding the external (verbal and nonverbal) behaviors and internal stress indicators (e.g., neurosensory) of its human counterpart. Development of an AI companion is especially challenging when working with members of a vulnerable population whose facial expressions and vocalizations do not always match those of neurotypical adolescents.

A current issue with this type of research is that training datasets are populated by neurotypical subjects and thus are potentially not representative of the population this project is intended to serve, adolescents with ASD. The desired outcomes from the project are novel deep learning algorithms to address the shortcomings of the state-of-the-art multimodal Facial Expression Recognition algorithms currently designed for neurotypical adolescents.

The project may also aid self-regulation of ASD adolescents when faced with cyberbullying and attempts to invade privacy as well as other social situations. It should also lead to an understanding of how this AI intervention strengthens the individual’s knowledge of technology and their affinity for cybersecurity careers.

This project is supported by the Secure and Trustworthy Cyberspace (SaTC) program, which funds proposals that address cybersecurity and privacy, and in this case specifically cybersecurity education. The SaTC program aligns with the Federal Cybersecurity Research and Development Strategic Plan and the National Privacy Research Strategy to protect and preserve the growing social and economic benefits of cyber systems while ensuring security and privacy.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.