Security

Machine Learning for IoT Security – with Dr Bob Baxley of Bastille

Radhika Madhavan
avatar

Radhika Madhavan covers AI innovation at Emerj. Radhika previously worked in content marketing at three technology firms, and graduated from Sri Krishna College Of Engineering And Technology with a degree in Information Technology.

Machine Learning for IoT Security - with Dr Bob Baxley of Bastille

This week on AI in Industry, we explore IoT security with Bob Baxley (Chief Engineer at Bastille). This includes information on how different IoT security is compared to infosec, the unique challenges IoT security presents (for detecting and scanning wireless network traffic that runs on various protocols and for classifying types of cyberthreats), what the future of IoT security might look like, and how deep learning and machine learning tools can be used to better classify and detect threats and attacks in the cyberspace.

Guest: Dr. Bob Baxley, Chief Engineer at Bastille

Expertise: Wireless security, data science, machine learning, software defined radio

Brief recognition: Bob Baxley is a Chief Engineer at Bastille Networks and has been with the company since 2014. He received his doctorate degree in Electrical Engineering from Georgia Institute of Technology, where he served as Adjunct Faculty for over four years.

Big Idea:

One of the biggest challenges in the IoT network security is classifying the type of data to look for anomalies, which are significant deviations in the way data usually behaves over the network. For enterprises and businesses, this poses a huge threat because it means their data on an IoT device–that might be running on any of the several non-standardized wireless protocols–is not secure.

Bob walks us through how this problem is currently tackled, and how IoT security could be improved in the future using deep learning and machine learning applications, such as TensorFlow. Below are some of the important points Bob covers in this particular section:

  • Machine-learning tools are already getting more accurate at classifying image categories with only a 3% error-rate as opposed to human error-rate of 5-10%. This is because they get trained on copious amounts of data and become better at identifying categories over time. In the near future, this capability could be extended to the cyberspace for classifying data threats.
  • Currently, there is a significant lack of labeled data in the cyberspace from which the machine learning applications can learn. Therefore, identifying anomalies and monitoring cybersecurity involve hours of manual work and effort.

Bob explains how the learnings from image classification can be applied to cybersecurity and how machine learning tools can better identify anomalies in IoT network traffic that could constitute as threats.

Interview Highlights on IoT Security

The main questions Bob answered on IoT security are listed below. Listeners can use the embedded podcast player (at the top of this post) to jump ahead to sections they might be interested in:

  • (2:00) How is IoT security different than infosec?
  • (4:17) What are the unique challenges when it comes to IoT network security?
  • (11:09) Future of IoT security: how can machine learning applications like TensorFlow help in improving cybersecurity?
  • (24:44) Can machine learning applications reduce manual intervention in cyberspace in the near future?

Subscribe to our AI in Industry Podcast with your favorite podcast service:

 

Header image credit: engie.be

Related Posts

Stay Ahead of the AI Curve

Discover the critical AI trends and applications that separate winners from losers in the future of business.

Sign up for the 'AI Advantage' newsletter:

Subscribe
Related posts (5)
Sumo Logic CTO - How Machine Learning Shines Light on Business Blind Spots 2
Business intelligence and analytics

Sumo Logic CTO – How Machine Learning Shines Light on Business Blind Spots

Episode Summary: CTO and Co-founder of Sumo Logic Christian Beedgen gives his take on how to glean return on investment from applying machine learning to companies. There are no easy answers, but Beedgen boils down simple concepts for thinking about humans thinking through causation, machines working out correlations, and how the combination of the two can glean us better ideas and get to answers faster than humans can do alone.
Machine Learning Cyber Security May Help Speed Response to Hack Attacks 2
Security

Machine Learning Cyber Security May Help Speed Response to Hack Attacks

Episode Summary: In this week’s episode, I speak with Igor Baikalov, chief scientist at cybersecurity company Securonix, about the trends in data security and where security itself has had to take a step up in the last five years. Igor touches on major meta-trends that have forced data security to advance, as well as what has made AI and machine learning a ‘requirement’ of modern data security strategy, something that has changed significantly in the last decade. Igor sheds light on these issues and likely future trends in cybersecurity over the next five to 10 years.
Network Intrusion Detection Using Machine Learning 4
Security

Network Intrusion Detection Using Machine Learning

Episode Summary: When Google’s DeepMind won against one of the best modern Go champions, is used multiple AI approaches and exposed gaps in some individual strategies. This even has shed more light on AI, but also on the utility in combining approaches to AI for individual problems. Data security is one of these problem areas where multiple AI approaches is being used to make our information safer. Dr. Sal Stolfo has been a professor at Columbia in Computer Science since 1972 and is now also the CEO of Allure Security, with a focus on engineering network intrusion detection solutions using AI applications. In this episode, Stolfo talks about the various styles of AI and statical methods that have been and are being used to detect malicious activity, as well as how he believes the future of security is going to have to adapt as increasing amounts of data become available.
Google Algorithm Disrupts Medical Field, Intel Launches Automated Driving Group, and More - This Week in Artificial Intelligence 12-02-16 1
Business intelligence and analytics
Finance
Data management

Crowdsourcing a Machine Learning Hedge Fund

Episode Summary: Crowdsourcing is a relatively common term in technical vernacular today. Even if you're not a self-identified "techie", you may very may well have leveraged crowdsourcing in journalism, the sciences, public policy, or elsewhere. One area in which this concept hasn’t really taken off is in finance and hedge funds. In this episode, we speak with Numerai Founder Richard Craib, whose company is crowdsourcing a machine learning hedge fund. Their model is based on pooling data science talent from all over the world and using "anonymous" models to train financial data. These models compete against one another, and the winning models' creators are rewarded in bitcoin - a process based entirely on encryption and anonymity. Craib speaks about his overarching vision for the company, and also delves into his thoughts on the past, present, and future of AI applications in finance.
 
Network Intrusion Detection Using Machine Learning 3
Data management

MuleSoft’s CTO Envisions Connected Machine Learning Network

Episode Summary: This episode's guest is Uri Sarid, PhD, CTO for MuleSoft, Inc. Sarid speaks about where he believes the future of machine learning (ML) applications in industry might go - he thinks applications might stay small and niche-based, and will develop based on how well each serves its individual purposes. He also gives his perspective on how companies may adapt to deal with these disparate ML technologies, and expands on his belief that finding ways to connect technologies will be an important path in the development of machine learning applications and platforms across industries.
© 2024 Emerj Artificial Intelligence Research. All rights reserved.