Josephine P
- Research Program Mentor

PhD candidate at Columbia University

Expertise

Computer Science, Security and Machine Learning, AI, Accessibilty in Computer Science, HCI, Robotics, Art- film photography, visual arts

Bio

Hi, I'm Josephine! I'm a first year PhD Student at Columbia University. These days I'm motivated by the growing application of ML into our daily lives and the security risks accompanying it. Many systems rely on personal data to improve user experiences, such as face/voice recognition, but potentially exposes our data to malicious attacks (such as deepfakes!). I'm exploring how we can mitigate these attacks by improving the robustness of ML algorithms and protecting the privacy of personally identifiable information (PII). In my free time I love to be outdoors- hiking, camping, sailing, playing tennis and going for walks or bike rides in the city! I also enjoy cooking with friends, hanging out with my cat and various arts n crafts. I recently started teaching at a Ceramics Studio and try to find time to create things in my school's fabrication lab.

Project ideas

Project ideas are meant to help inspire student thinking about their own project. Students are in the driver seat of their research and are free to use any or none of the ideas shared by their mentors.

Exploratory Research Paper

Deepfakes, Voice mimic attacks, etc. are all made possible because of the leakage of personal information. But what type of data is needed to implement such an attack? How much information about an individual is needed? More importantly, where does this data come from? Project formation: chose a motivating attack (i.e. voice cloning, face cloning, motion cloning) and perform a literature review to determine what type of data is needed to mount the attack. Then explore where this information can be collected from: examine common application/systems that might collect this type of data and investigate its security/privacy policies. This project is most likely thought of in terms of facial recognition, where photos can be easily scraped from public profile photos on the web. But where might systems collect voice data? motion data (i.e mimicking an individual's walk)? Even data capturing how a person speaks (what words, phrases, etc. represent them and their writing style) In this project you will write a research paper identifying and exploring in-depth an area of current privacy concern. You will gain skills reading top-tier research papers and discovering how to *independently* find a research are of interest. This project can easily be expanded in scope for future projects.

Coding skills

python, C, tensorflow, pytorch, adequacy in most common languages

Languages I know

French, beginner/intermediate

Teaching experience

Designed and taught 2 week intensive summer courses in C programming for advanced high school students. Mentored middle school-high school female students in weekend hackathons (Volunteer). Mentor undergraduate SWE students at Columbia University in research/career fields (Volunteer). Flute teacher for the South Side Free Music Program in Chicago (Volunteer).