Automated Email Generation for Targeted Attacks


listen on castbox.fmlisten on google podcastslisten on player.fmlisten on pocketcastslisten on podcast addictlisten on tuninlisten on Amazon Musiclisten on Stitcher

--:--
--:--


2022-10-31

Automated Email Generation for Targeted Attacks

Avisha Das, a post-doctoral scholar at the University of Texas Health Center at Houston, joins us today. Avisha’s research studies phishing attacks and natural language models in biomedical applications.

She started the conversation by discussing what phishing was, and how attackers solicit classified information from unsuspecting victims. She differentiated how phishing was different from spam emails and the effort email service providers are making to clamp down on it. She then discussed a different kind of phishing called spear phishing.

Avisha revealed how attackers, after gaining access to a victim’s PC, use machine-learning techniques to mirror the email patterns of other co-workers. This way they are more difficult to detect. In her research, she sought to replicate this email generation pipeline. She discussed some of the roadblocks she faced in generating new emails from the email dataset.

Avisha discussed how she first discovered machine learning could learn how humans write. At the time, there was no BERT or GPT-3. She explained why the phishing email problem is still widely unsolved. She also discussed possible ways one can identify phishing emails and take caution. Avisha then spoke on how fraudulent websites train their chatbots with machine learning to get information from people.

Going forward, Avisha explained how chatbots can be positively used for health and emotional therapies. She discussed the possibility of applying generative language models for cognitive therapies as well.

Avisha then spoke about the challenge of language models being a black box. She discussed ways researchers are handling this black box problem. But still, she mentioned how there is still a long way to go to pass the Turing test. Concluding, she spoke about the projection of this field in the next 5 years. You can learn more about Avisha’s work from her website, or follower her on LinkedIn. You may also check out her Google Scholar page to stay up to speed with her research.

Avisha Das

Avisha Das is a Postdoctoral Scholar at the School of Biomedical Informatics, University of Texas Health Science Center in Houston. Her research focuses on developing automated natural language understanding and generation models, targeted towards short text generation or task-based conversation. She received her Ph.D. from University of Houston in 2020, where her thesis focused on the feasibility of generating automated social engineering attacks at scale and the likelihood of humans falling victim to such threats. She is currently a BIG-TCR CPRIT Postdoctoral Fellow at UTHealth and working on mining biomedical knowledge for automated content distillation in cancer research. Her research interests include Biomedical Knowledge Mining, Natural Language Generation, and Security Analytics.


Thanks to our sponsors for their support

Astrato is a modern BI and analytics platform built for the Snowflake Data Cloud. A next-generation live query data visualization and analytics solution, empowering everyone to make live data decisions.
https://astrato.io