Medium

Smart and connected devices, also known as Internet of Things (IoT) devices, are now an integral part of our daily lives. These devices are found in cars, phones, watches, appliances, home security systems, and in critical applications, such as utilities and in the biomedical industry. The convenience provided by IoT devices comes with unique security and privacy concerns. Because of the shortened time-to-market and the fierce competition among companies, security has not been treated as a priority in these devices.

Adversaries are outpacing developers in the race to find program vulnerabilities. Where programmers have to find all potential software flaws in their programs and determine whether they are exploitable across all deployments to prevent vulnerabilities, adversaries need only find one software flaw that enables them to achieve their goals in any one deployment. Current techniques to rid programs of vulnerabilities cannot find all such flaws due to the complexity of modern software and their deployments.

Smart and connected devices, also known as Internet of Things (IoT) devices, are now an integral part of our daily lives. These devices are found in cars, phones, watches, appliances, home security systems, and in critical applications, such as utilities and in the biomedical industry. The convenience provided by IoT devices comes with unique security and privacy concerns. Because of the shortened time-to-market and the fierce competition among companies, security has not been treated as a priority in these devices.

Machine learning techniques, particularly deep neural networks, are increasingly integrated into safety and security-critical applications such as autonomous driving, precision health care, intrusion detection, malware detection, and spam filtering. A number of studies have shown that these models can be vulnerable to adversarial evasion attacks where the attacker makes small, carefully crafted changes to normal examples in order to trick the model into making incorrect decisions.

As software now pervades nearly every aspect of modern life, securing software is widely acknowledged as a critical problem. Although significant effort has gone into identifying flaws in software, as well as developing tools, libraries, and processes for detecting and mitigating these flaws during software development and maintenance, security problems remain pervasive.