Cătălina Cangea
My website has moved to a new location :-).
I am a second-year PhD student within the Artificial Intelligence Group of the Department of Computer Science and Technology (formerly known as the Computer Laboratory). My research focuses on learning from complex-structured data in multimodal and relational settings. I am supervised by Pietro Liò, with Mateja Jamnik as my second adviser, and affiliated with King's College. I was the chair of women@CL in 2018-19 and am deputy chair for 2019-20.
I graduated from the Computer Science Tripos in 2016 with First Class and hold an Advanced Computer Science MPhil degree with Distinction, as of July 2017.
Publications
Purves, C., Cangea, C. and Veličković, P. (2019) The PlayStation Reinforcement Learning Environment (PSXLE). Deep Reinforcement Learning Workshop at the Thirty-third Conference on Neural Information Processing Systems (NeurIPS 2019).
Cangea, C., Belilovsky, E., Liò, P. and Courville, A. (2019) VideoNavQA: Bridging the Gap between Visual and Embodied Question Answering. The 30th British Machine Vision Conference (BMVC 2019). [arXiv] [poster] [code]
Opolka, F.*, Solomon, A.*, Cangea, C., Veličković, P., Liò, P. and Hjelm, D. (2019) Spatio-Temporal Deep Graph Infomax. Representation Learning on Graphs and Manifolds (RLGM) Workshop at the 7th International Conference on Learning Representations (ICLR 2019). [arXiv]
Cangea, C., Grauslys, A., Liò, P. and Falciani, F. (2018) Structure-Based Networks for Drug Validation. Machine Learning for Health (ML4H) Workshop at the 32nd Annual Conference on Neural Information Processing Systems (NeurIPS 2018). A subsequent study was presented at the AI for Social Good (AISG) Workshop at the 7th International Conference on Learning Representations (ICLR 2019). [arXiv]
Cangea, C.*, Veličković, P.*, Jovanović, N., Kipf, T. and Liò, P. (2018) Towards Sparse Hierarchical Graph Classifiers. Workshop on Relational Representation Learning (R2L) at the 32nd Annual Conference on Neural Information Processing Systems (NeurIPS 2018). [arXiv]
Cangea, C., Veličković, P. and Liò, P. (2017) XFlow: 1D-2D Cross-modal Deep Neural Networks for Audiovisual Classification. Workshop on Computational Models for Crossmodal Learning (CMCML) at The 7th Joint IEEE International Conference on Development and Learning and on Epigenetic Robotics (IEEE ICDL-EPIROB 2017). [arXiv] [poster] [code]
This work was also presented at the ARM Research Summit 2017, in the poster session.
Research community involvement
Reviewer for the following venues:
- Visually Grounded Interaction and Language (ViGIL), workshop at NeurIPS 2019.
- Graph Representation Learning, workshop at NeurIPS 2019.
- Learning and Reasoning with Graph-Structured Data, workshop at ICML 2019.
- Representation Learning on Graphs and Manifolds, workshop at ICLR 2019.
- 13th Women in Machine Learning Workshop (WiML 2018), co-located with NeurIPS 2018.
Work experience and other projects
PhD internships:
- Summer 2019, AI Residency at X (including a detour at DLRLSS 2019 in Edmonton!)
- Summer 2018, Montreal Institute for Learning Algorithms, under the supervision of Aaron Courville.
Software engineering internships:
- July-September 2016, Facebook London, LogDevice team. I optimised client operations on a RocksDB database and implemented a new API required by another team in Facebook.
- July-September 2015, Facebook New York, iOS Product Infrastructure Team. I worked towards delivering a better experience for users of the Facebook iOS app. My project aimed to reduce the time taken to load content close to the area currently being viewed on screen, by improving the prioritization system for network requests.
- June-September 2014, Google Zurich, YouTube Uploads team. I added processing progress for video uploads on several YouTube pages, as the Upload page was the only one displaying this information.
I also took part in Hack Cambridge, winning third place in 2017. We showed that you can trick a face recognition system quite easily, if you've got access to the neural network architecture!
Talks and presentations
Hierarchical Music Generation. AI Residency Journal Club, 21 August 2019. X, the moonshot factory, Mountain View, CA.
VideoNavQA: Bridging the Gap between Visual and Embodied Question Answering. AI Residency Colloquium, 7 August 2019. X, the moonshot factory, Mountain View, CA.
Towards Sparse Hierarchical Graph Classifiers. (Remote talk) Graph Representation Reading Group, 15 November 2018 (Also presented at the Computational Biology/Artificial Intelligence Group Meeting, University of Cambridge, 13 November 2018). Mila, University of Montreal
Graph Convolutional Neural Networks for Web-Scale Recommender Systems. Graph Representation Reading Group, 20 September 2018. Mila, University of Montreal
NerveNet: Learning Structured Policy with Graph Neural Networks. Graph Representation Reading Group, 15 August 2018. Mila, University of Montreal
Cross-modal Data Integration Techniques. PropagAgeing Workshop Meeting, 7 February 2018 (Also presented at the Women@CL TalkLet, 25 May 2018). Department of Computer Science, University of Cambridge
Cross-modality in Deep Learning for Audiovisual Classification. Computational Biology Group Meeting, 22 January 2018. Department of Computer Science, University of Cambridge
Academic involvement
Undergraduate admissions interviews:
- 5 days, December 2017: King's College.
- 1 day, December 2016: Murray Edwards College.
(Co-supervisor) MPhil in Advanced Computer Science research projects:
- Felix Opolka, Representation Learning for Spatio-Temporal Graphs (2018-19) (85/100).
- Aaron Solomon, Dynamic Temporal Analysis for Graph Structured Data (2018-19).
(Official supervisor) Computer Science Tripos Part II projects:
- Carlos Purves, The PlayStation Reinforcement Learning Environment (2018-19) (80/100).
- Andrew Wells, Deep Learning for Music Recommendation (2017-18) (76/100).
Undergraduate courses for Murray Edwards, King's, and Queens' Colleges:
- Artificial Intelligence (Easter '17, '19, '20)
- Databases (Michaelmas '17, '18)
- Discrete Mathematics (Michaelmas '17, '18, Lent '18, '19)
- Foundations of Computer Science (Michaelmas '16, '17, '18)
- Logic and Proof (Lent '17, '20)
- Machine Learning and Real-world Data (Lent '18)
External teaching
I am a Machine Learning Teaching Fellow for Cambridge Spark. I teach the Neural Networks module from the Applied Data Science London Bootcamp to industry professionals.
In my "free time"
In 2016, I took up rowing as a member of Darwin College Boat Club until October 2017, racing in Bumps as part of the Women's Second Boat. I'm currently training with the Women's First Boat in King's College Boat Club.
My relentless lifelong passion is music - I play piano and guitar, having been part of a couple of music bands (keyboards, backing vocals, guitar, vocals...) ever since early 2017. We've performed at pub gigs, open mic nights, at Darwin May Ball in 2017 and St John's May Ball in 2018.
Contact
Address:
Cătălina Cangea
Office FE14
Artificial Intelligence Group
Department of Computer Science and Technology
University of Cambridge
15 JJ Thomson Avenue
Cambridge CB3 0FD
United Kingdom
Email: Catalina [dot] Cangea [at] cst [dot] cam [dot] ac [dot] uk