My name is Huang Xiao. I spent some wonderful time in [research](https://scholar.google.com/citations?user=vEep9sYAAAAJ&hl=en) and acquired my Ph.D in computer science that focused on adversarial machine learning. I have worked in different industries, mainly Cybersecurity, but also with experience in automotive and financing. I love challenges and solve problems using scientific approaches. I built this website as part of my key principles of `Research`,`Engineering`,`Create` and `Impact`. I love research, especially on those hard and challenging problems, I also believe in bringing scientific results into real life, which requires a fair amount of engineering work done to make sure we work around those *assumptions*. We realise that sometimes it is impossible to achieve that goal, however, we aim to create real value and build tangible products or services that make societal impact. ## My interests I start my passion with machine learning since 2009 during my master study at [TUM](https://www.tum.de/en/studies/degree-programs/detail/informatics-master-of-science-msc). Ever since then I am interested at machine learning in general. During my Ph.D I focus on the topic [[Introduction|adversarial machine learning]], and I still remain interest at the same topic. At that time, adversarial ML is not getting enough attention, since most researchers are busy with proposing new learning methods to increase model generalisation, and yes - this still remains at the core of machine learning. With the landscape being shifted these years, many of us realised that model generalisation needs to be expanded to cope with adversarial attempts. The problem is far from being solved, therefore I'd spend more time in this. In general, I'd like to be able to answer some key questions: 1. Given a model / algorithm, can we find its vulnerability i.e., adversarial samples, prompts at what cost? 2. What countermeasures are there to safeguard the vulnerabilities, can we propose learning algorithms that are more robust against both noises and adversarial manipulation? 3. What we can do to standardise the security control around a model, so that we can be confident of deploying it in production? The ultimate goal of AGI (arguably well defined) is not going to bypass these questions. If you work in the same areas or are interested at these topics, feel free to [contact me](mailto:[email protected]), I am open to all kinds of collaboration. ## Short bio. #### 🎓Education --- > **2016 - 2017** > Stanford University > Visiting scholar in HCI Group, Computers science dept. > **2012 - 2016** > Technical University of Munich > Ph.D, Computer science, supervised by [Prof. Dr. Claudia Eckert](https://de.wikipedia.org/wiki/Claudia_Eckert) > Dissertation: [Adversarial and Secure Machine Learning](https://www.semanticscholar.org/paper/Adversarial-and-Secure-Machine-Learning-Xiao/2b3756f230463463b134e873c079ea94d103f6a8.) > **2009 - 2011** > Technical University of Munich > M.Sc. Computer science > Thesis: Master: Structure Learning of Copula Bayesian Networks. > **2002 - 2007** > Tongji University > B.Sc. Computer science #### 💼 Work Experience ---- > **2019.07 - Now** > HSBC Plc, London United Kingdom > Principal AI researcher/engineer > **2018.07 - 2019.07** > Bosch Center for Artificial Intelligence, Renningen, Germany > Research scientist > **2016.10 - 2018.06** > Fraunhofer AISEC Institute, Munich, Germany > Head of Research Group > **2015.10 - 2016.01** > Alibaba Group Inc., Hangzhou, China > Visiting Researcher > **2012.03 - 2012.09** > Fraunhofer AISEC Institute, Munich, Germany > Research Intern > **2009.10 - 2011.03** > Technical University of Munich, Germany > Assistant Researcher > **2007.06 - 2008.07** > Pioneer Suntec Electronic Technology, Shanghai, China > Software Engineer #### 📚 Publications ---- - Bojan Kolosnjaji, Huang Xiao, Peng Xu, Apostolis Zarras. Artificial Intelligence for Cybersecurity: Develop AI approaches to solve cybersecurity problems in your organization. Packt Publishing Ltd, 2024.10. - __Huang Xiao__, George Webster, Bojan Kolosnaji, Andrew Carney. **Network inventory management and anomaly detection system**. Application No.: 17/945.711. US Patent and Trade Office. Projected publication date: 03/21/2024. - Huang Xiao, Michael Herman. **Computer-Implemented Method of and Apparatus for Training a Neural Network**. Patent application [EP20190175574 20190521], 2020.11.25. - __Huang Xiao__, M. Lampacrescia. **Method for Calibrating a Multi-sensor System Using an Artificial Neural Network**. Patent application [US202016879335 20200520], 2020.12.03. - __Huang Xiao__, M. Herman, J. Wagner, S. Ziesche, J. Etesami, TH. Linh. **Wasserstein adversarial imitation learning**. arXiv preprint arXiv:1906.08113. June. 2019. - Felix Fischer, __Huang Xiao__, Ching-Yu Kao, Yannick Stachelscheid, Benjamin Johnson, Danial Razar. **Stack Overflow Considered Helpful! Deep Learning Security Nudges Towards Stronger Cryptography**. 28th Usenix security symposium 2019, accepted. - Xiao, Huang. **Adversarial and Secure Machine Learning**. Dissertation. Technical University of Munich, Oct. 2017. - F. Fischer, Konstantin B., __Huang Xiao__, Christian S., Yasemin A., Michael B., Sascha F.. **Stack Overflow Considered Harmful? - The Impact of Copy&Paste on Android Application Security**. In IEEE Conf. on Security \& Privacy, San Jose, CA, USA, 2017 (accept rate 13\%) . - Xiao, Huang, Battista Biggio, Gavin Brown, Giorgio Fumera, Claudia Eckert, and Fabio Roli. **Is Feature Selection Secure against Training Data Poisoning?**. In ICML'15, Lille, France, July 2015. - Xiao, Huang, Battista Biggio, Blaine Nelson, Han Xiao, Claudia Eckert, and Fabio Roli. **Support Vector Machines under Adversarial Label Contamination**. Journal of Neurocomputing, Special Issue on Advances in Learning with Label Noise, August 2014. - Xiao, Huang, and Claudia Eckert, **Indicative Support Vector Clustering with its Application on Anomaly Detection**. In IEEE 12th International Conference on Machine Learning and Applications (ICMLA'13), Miami, Florida, December 2013 - Xiao, Han, Xiao, Huang, and Claudia Eckert. **Learning from Multiple Observers with Unknown Expertise**. In Proceedings of 17th Pacific-Asia Conference on Knowledge Discovery and Data Mining, Gold Coast, Australia, Springer, April 2013. - Xiao, Huang, Xiao, Han, and Claudia Eckert. **OPARS: Objective Photo Aesthetics Ranking System**. In 34th European Conference on Information Retrieval (ECIR'13), Moscow, Russia, March 2013. - Xiao, Han, Xiao, Huang, and Claudia Eckert. **Adversarial Label Flips Attack on Support Vector Machines**. In 20th European Conference on Artificial Intelligence (ECAI), Montepellier, France, August 2012. - Xiao, Huang. **Structure Learning in Copula Bayesian Networks**. Master thesis, Technische Universität München Press, November 2011 - Akram, Hasan Ibne, Colin de la Higuera, Xiao, Huang, and Claudia Eckert. **Grammatical Inference Algorithms in MATLAB**. In ICGI 2010: Proceedings of the 10th International Colloquium on Grammatical Inference, Springer-Verlag, Valencia, Spain, 2010.