My PhD research is focused on privacy-preserving natural language processing.
Selected Academic Publications
Thaine, P., Penn, G. (2021). The Chinese Remainder Theorem for Compact, Task-Precise, Efficient and Secure Word Embeddings. In Proceedings of the 16th conference of the European Chapter of the Association for Computational Linguistics, EACL 2021.
Thaine, P., Penn, G. (2020). Vec2int: Applications of the Chinese Remainder Theorem in Word Embedding Compression and Arithmetic (Poster). Vector Institute Natural Language Processing Symposium, September 15 and 16, 2020.
Thaine, P., Penn, G. (2020). Reasoning about unstructured data de-identification. In Journal of Data Protection and Privacy, Vol. 3, No. 3.
Thaine, P., Penn, G. (2019). Vocalic and Consonantal Grapheme Classification through Spectral Decomposition. In Graphemics in the 21st Century, Proceedings of Grapholinguistics and Its Applications, Vol. 1.
Thaine, P., Penn, G. (2019). Extracting Bark-Frequency Cepstral Coefficients from Encrypted Signals. In Proceedings of INTERSPEECH, Graz, Austria. pdf blog
Thaine, P., Gorbunov, S., Penn, G. (2019). Efficient Evaluation of Activation Functions over Encrypted Data. In Proceedings of the 2nd Deep Learning and Security Workshop, 40th IEEE Symposium on Security and Privacy, San Francisco, USA. pdf
Thaine, P., Penn, G. (2019). Perfectly Privacy-Preserving AI: What is it and how do we achieve it? (Poster). Canada-United Kingdom Symposium on Ethics in Artificial Intelligence, EIAI-2019, Ottawa, Canada.
Thaine, P., Penn, G. (2019). Privacy-Preserving Character Language Modelling. In Proceedings of the Privacy-Enhancing Artificial Intelligence and Language Technologies AAAI Spring Symposium, PAL 2019, Stanford University, Palo Alto, USA. pdf
Thaine, P., Penn, G. (2018). Vocalic and Consonantal Grapheme Classification through Spectral Decomposition. In Graphemics in the 21st Century: Proceedings of the 2018 Conference (Grapholinguistics and Its Applications), Fluxus Editions. pdf amazon code
Thaine, P., Penn, G. (2017). Vowel and Consonant Classification through Spectral Decomposition. Proceedings of the First Workshop on Subword and Character Level Models in NLP, EMNLP, 2017. pdf code
Sultanum, N., Thaine, P., Brudno, M., Glueck, M., Wigdor, D., Chevalier, F. (2017, October). MedStory: Unlocking the Qualitative Power of Medical Narratives. In Proceedings of 8th Workshop on Visual Analytics in Healthcare (VAHC). pdf
Thaine P., Penn, G. (2016). A Survey of the State-of-the-Art in Acoustic Forensics. In Proceedings of the 18th Interpol Forensic Science Symposium, Lyon, France. 2016.
Rudzicz, F., Frydenlund, A., Robertson, S., Thaine, P. (2016, March). Acoustic-articulatory relationships and inversion in sum-product and deep-belief networks. In Speech Communication. pdf
Thaine, P., Penn, G. (2015, July). Writing Systems (Poster). Linguistic Society of America Linguistic Summer Institute, Chicago, IL.
Talks and Panels
Dealing with Personal Data using AI (December 15, 2021), at the Better Ethics and Consumer OUrcomes Network's Fireside Chat.
Privacy-Enhancing Technologies in AI Security (december 1, 2021), at O'Reilly Media's AI Superstream Series: Securing AI.
Panel: Big Data and Analytics Strategy at the Heart of Cybersecurity and Privacy (November 18, 2021), at Toronto Machine Learning Summit.
Panel: Demystifying the De-identification of Data (November 16, 2021), at The Innovation Game: Adopting RegTech in a Digital Age, Canadian Regulatory Technology Association.
Panel: Can voices be anonymised? (November 2, 2021), Lorentz Workshop on Speech as Personable Identifiable Information.
The Latest Advances in Privacy-Preserving NLP (September 21, 2021), Toronto Machine Leaning Summit on NLP.
Privacy Preserving Synthetic Data in AI/ML - A Mirage. (June 2, 2021), Privacy Symposium 2021 (Infosys - IAPP).
Efficient Evaluation of Activation Functions over Encrypted Data (January 15, 2021), UofT AI Conference.
Private-Preserving Machine Learning (December 3, 2020), MLOps: Production and Engineering Vancouver 2020.
Cybersecurity and Privacy: Complements for a more secure Internet (November 25, 2020), Keynote talk at Vector Institute Endless Summer School (ESS).
Panel moderator for The Role of ML in Climate Change (November 18, 2020), at Toronto Machine Learning Summit.
Privacy in Deployment (October 16, 2020), 2020 USENIX Conference on Privacy Engineering Practice and Respect (PEPR'20).
A Practical Guide to Privacy-Preserving Machine Learning (November 12, 2020), EVOKE CASCON 2020.
Privacy in Production (June 30, 2020), Canada AI/ML, Data Science and Engineering Digital Meetup.
Privacy-Preserving Machine Learning (June 18, 2020), MLOps: Production and Engineering World.
Privacy-Preserving Machine Learning: A Practical Overview (June 10, 2020), Vector Institute Endless Summer School (ESS).
An Overview of the Problem of Perfectly Privacy-Preserving AI (June 8, 2020), Future of Privacy Forum AI Working Group.
Privacy-Preserving Natural Language Processing Using Homomorphic Encryption (2019), National Research Council of Canada, Ottawa, CA.
Privacy-Preserving Natural Language Processing Using Homomorphic Encryption (2019), Borealis AI, Toronto, CA.
Perfectly Privacy-Preserving AI: What is it and how do we achieve it? (2019), Identity, Privacy, and Security Institute, Toronto.
Privacy-Preserving Natural Language Processing (2018), Vector Institute for Artificial Intelligence, Toronto, CA.
Vowel and Consonant Classification through Spectral Decomposition (2017), National Research Council of Canada, Ottawa, CA.
EY Women in Tech Award (October 2020).
Satchu Prize in recognition of outstanding performance in the program and a demonstrated potential to lead Canada's next generation of high impact entrepreneurs (October 2020).
RBC Graduate Fellowship in recognition of excellence in research and interest in commercialization (2019 – 2021).
NSERC PGSD in recognition of excellence in research and academic performance (2017 – 2020).
Beatrice "Trixie" Worsley Graduate Scholarship in Computer Science (November 2017).
Ontario Graduate Scholarship in recognition of excellence in research and academic performance (2016 – 2017).
Service to the Community
Co-Organizer, PrivateNLP Workshop (June 2021), co-located with NAACL 2021.
Reviewer, Journal of Medical Internet Research (August 2020).
Program Committee Member, EMNLP 2020.
Co-Organizer, PrivateNLP Workshop (November 20 2020), co-located with EMNLP 2020.
Co-Organizer, PrivateNLP Workshop (February 7 2020), co-located with the 13th ACM International WSDM Conference (WSDM 2020).
Program Committee Member, 2019 Privacy-Enhancing Artificial Intelligence and Language Technologies AAAI Spring Symposium.
Program Committee Member in the area of Ethics, Bias, and Fairness, NAACL-HLT 2019.
Program Committee Member, Eighth Joint Conference on Lexical and Computational Semantics (SEM 2019).
Reviewer, Computational Linguistics Journal (November 2017, April 2017, July 2019).
Reviewer, Computational Intelligence Journal (March 2018).
Reviewer, Twenty-Ninth AAAI Conference on Artificial Intelligence (AAAI-15).
The Last Watchdog and Security Boulevard Guest Essay: How stricter data privacy laws have redefined the ‘filing’ of our personal data, guest essay, November 4th, 2021.
Unite.ai, An interview with Patricia Thaine, CEO at Private AI, interview, October 19, 2021.
TechTarget, FTC pursues AI regulation, bans biased algorithms, commentary, October 19, 2021.
Betanews, Industry leaders comment on Cybersecurity Awareness Month, commentary, October, 2021.
Security Magazine, October marks Cybersecurity Awareness Month: Security experts comment on where efforts should be focused, commentary, October 1, 2021.
insideBigData, Heard on the Street - 9/27/2021, commentary, September 27, 2021.
Press Release, Private AI secures $3.15M seed round to streamline privacy compliance for enterprises, announcement, September 15, 2021. Published in Fortune's Term Sheet Newsletter, BetaKit, Cyberwire, MarTech Series, AI Techpark, TMCNet, Private Capital Journal, Enterprise AI, AiThority, Dark Reading, Fintech Global, CX Today, Venture Capital Journal, insideBigData, and Analytics Insight.
Threat Technology, Private AI Delivers companies an easy-to-deploy way to redact sensitive data on the files they share, interview, September 9, 2021.
Invest Ontario, 10 cybersecurity companies to watch in 2021, August 6, 2021.
The Local Maximum, Episode 181 - Redacting your Secrets, July 19, 2021.
Threat Technology, These are the Top Cyber Security Companies in Toronto (2021), January 23, 2021.
Forbes, This Startup Founder Sees A Data Privacy Reckoning On The Horizon, interview, January 21, 2021.
Founded by Women, featured in book, January 7, 2021.
Privacy by Design Lab (Japan), Privacy Talk with Patricia Thaine, Co-Founder and CEO of Private AI, interview, January 2, 2021.
Datacast, Episode 42: Privacy-Preserving Natural Language Processing with Patricia Thaine, podcast, September 11, 2020.
Anonymized data is useless: fact or fiction?, August 17, 2021.
Data Anonymization: Perspectives from a Former Skeptic, June 4, 2021.
Demystifying De-identification: Understanding key tech for data protection regulation compliance, April 6, 2021.
Cybersecurity and Privacy: Complements for a more secure Internet, December 14, 2020.
Privacy Enhancing Technologies Decision Tree (v2), October 18, 2020.
Perfectly Privacy-Preserving AI: What is it and how do we achieve it?, January 1, 2020.
Homommorphic Encryption for Beginners: A Practical Guide (Part 2: The Fourier Transform), September 3, 2019.
Differentially Private Natural Language Processing, January 28, 2019.
Homomorphic Encryption for Beginners: A Practical Guide (Part 1), December 26, 2018.
Why is Privacy-Preserving Natural Language Processing Important?, June 26, 2018.
A Brief Overview of Privacy-Preservinc Software Methods, May 22, 2018.