Home.

About me

Education

• PhD at University College London. (2018-Present)
  • Funded by EPSRC and Siemens Healthineers.
• MRes in Quantum Technologies at University College London. (2017-2018)
  • Awarded the MRes Prize for Outstanding Performance (joint top performance in my cohort).
  • Funded by EPSRC and Siemens Healthineers.
• MSci Physics and Physical Chemistry at University College London. (2013-2017)
  • Appeared on the Dean’s List 2017 (top 5% of graduating students).
  • Exchange year at Nanyang Technological University, Singapore. (2015-2016)

Experience

• Deep Learning Intern at Siemens Healthineers (July 2019 - January 2020)
• Deep Learning Intern at Imperial College London (June - August 2017)
• Database Intern at Quantemol (July - September 2016)

Teaching

• Teaching Assistant for Supervised Learning course (COMP0078). (Fall 2019)
  • Delivered tutorials, answering any questions that students had on the course material and problem sheets.
  • Developed code to identify duplicate PDF submissions, increasing the marking efficiency.
  • Coordinated the marking efforts for the second coursework.
• Demonstrator for Laboratory course (PHAS0007). (Fall 2018)
  • Assisted students with experiments and data analysis.
• Teaching Assistant for Scientific Communication course. (Fall 2017)
  • Provided detailed feedback on coursework (articles on scientific topics written for a general audience).

Research.

Publications & Preprints

1. Tse, L., Amadou, A. A., Georget, A., & Tuysuzoglu, A. (2021). PRANet: Point Cloud Registration with an Artificial Agent. ArXiv.
   Abstract Bibtex ArXiv
   Point cloud registration plays a critical role in a multitude of computer vision tasks, such as pose estimation and 3D localization. Recently, a plethora of deep learning methods were formulated that aim to tackle this problem. Most of these approaches find point or feature correspondences, from which the transformations are computed. We give a different perspective and frame the registration problem as a Markov Decision Process. Instead of directly searching for the transformation, the problem becomes one of finding a sequence of translation and rotation actions that is equivalent to this transformation. To this end, we propose an artificial agent trained end-to-end using deep supervised learning. In contrast to conventional reinforcement learning techniques, the observations are sampled i.i.d. and thus no experience replay buffer is required, resulting in a more streamlined training process. Experiments on ModelNet40 show results comparable or superior to the state of the art in the case of clean, noisy and partially visible datasets.

   @article{Tse2021, author = {Tse, Lisa and Amadou, Abdoul Aziz and Georget, Axen and Tuysuzoglu, Ahmet}, title = {PRANet: Point Cloud Registration with an Artificial Agent}, year = {2021}, journal = {ArXiv}, arxivid = {2109.11349} }

2. Herbster, M., Pasteris, S., & Tse, L. (2020). Online matrix completion with side information. Neural Information Processing Systems, 33.
   Abstract Bibtex PDF ArXiv
   We give an online algorithm and prove novel mistake and regret bounds for online binary matrix completion with side information. The bounds we prove are of the form O(D/gamma^2 ). The term 1/gamma^2 is analogous to the usual margin term in SVM (perceptron) bounds. More specifically, if we assume that there is some factorization of the underlying m x n matrix into PQ^T where the rows of P are interpreted as "classifiers" in R^d and the rows of Q as "instances" in R^d, then gamma is is the maximum (normalized) margin over all factorizations PQ^T consistent with the observed matrix. The quasi-dimension term D measures the quality of side information. In the presence of no side information, D = m + n. However, if the side information is predictive of the underlying factorization of the matrix, then in the best case, D in O(k + l) where k is the number of distinct row factors and l is the number of distinct column factors. We additionally provide a generalization of our algorithm to the inductive setting. In this setting, the side information is not specified in advance. The results are similar to the transductive setting but in the best case, the quasi-dimension D is now bounded by O(k^2 + l^2).

   @article{Herbster2020, archiveprefix = {arXiv}, arxivid = {1906.07255}, author = {Herbster, Mark and Pasteris, Stephen and Tse, Lisa}, eprint = {1906.07255}, link = {https://proceedings.neurips.cc/paper/2020/file/eb06b9db06012a7a4179b8f3cb5384d3-Paper.pdf}, issn = {23318422}, journal = {Neural Information Processing Systems}, number = {33}, title = {{Online matrix completion with side information}}, year = {2020} }

3. Herbster, M., Pasteris, S., & Tse, L. (2020). Online multitask learning with long-term memory. Neural Information Processing Systems, 33.
   Abstract Bibtex PDF ArXiv
   We introduce a novel online multitask setting. In this setting each task is partitioned into a sequence of segments that is unknown to the learner. Associated with each segment is a hypothesis from some hypothesis class. We give algorithms that are designed to exploit the scenario where there are many such segments but significantly fewer associated hypotheses. We prove regret bounds that hold for any segmentation of the tasks and any association of hypotheses to the segments. In the single-task setting this is equivalent to switching with long-term memory in the sense of [1]. We provide an algorithm that predicts on each trial in time linear in the number of hypotheses when the hypothesis class is finite. We also consider infinite hypothesis classes from reproducing kernel Hilbert spaces for which we give an algorithm whose per trial time complexity is cubic in the number of cumulative trials. In the single-task special case this is the first example of an efficient regret-bounded switching algorithm with long-term memory for a non-parametric hypothesis class.

   @article{Herbster2020a, archiveprefix = {arXiv}, arxivid = {2008.07055}, author = {Herbster, Mark and Pasteris, Stephen and Tse, Lisa}, eprint = {2008.07055}, issn = {23318422}, journal = {Neural Information Processing Systems}, month = aug, publisher = {arXiv}, title = {{Online multitask learning with long-term memory}}, volume = {33}, link = {https://papers.nips.cc/paper/2020/file/cdfa4c42f465a5a66871587c69fcfa34-Paper.pdf}, year = {2020} }

4. Tse, L., Mountney, P., Klein, P., & Severini, S. (2018). Graph Cut Segmentation Methods Revisited with a Quantum Algorithm. ArXiv.
   Abstract Bibtex ArXiv
   The design and performance of computer vision algorithms are greatly influenced by the hardware on which they are implemented. CPUs, multi-core CPUs, FPGAs and GPUs have inspired new algorithms and enabled existing ideas to be realized. This is notably the case with GPUs, which has significantly changed the landscape of computer vision research through deep learning. As the end of Moores law approaches, researchers and hardware manufacturers are exploring alternative hardware computing paradigms. Quantum computers are a very promising alternative and offer polynomial or even exponential speed-ups over conventional computing for some problems. This paper presents a novel approach to image segmentation that uses new quantum computing hardware. Segmentation is formulated as a graph cut problem that can be mapped to the quantum approximate optimization algorithm (QAOA). This algorithm can be implemented on current and near-term quantum computers. Encouraging results are presented on artificial and medical imaging data. This represents an important, practical step towards leveraging quantum computers for computer vision.

   @article{Tse2018, author = {Tse, Lisa and Mountney, Peter and Klein, Paul and Severini, Simone}, journal = {ArXiv}, title = {Graph Cut Segmentation Methods Revisited with a Quantum Algorithm}, year = {2018}, arxivid = {1812.03050} }

5. Tennyson, J., Rahimi, S., Hill, C., Tse, L., Vibhakar, A., Akello-Egwel, D., Brown, D. B., Dzarasova, A., Hamilton, J. R., Jaksch, D., Mohr, S., Wren-Little, K., Bruckmeier, J., Agarwal, A., Bartschat, K., Bogaerts, A., Booth, J.-P., Goeckner, M. J., Hassouni, K., … Yoon, J.-S. (2017). "QDB: A new database of plasma chemistries and reactions". Plasma Sources Science and Technology, 26(5). https://doi.org/10.1088/1361-6595/aa6669
   Abstract Bibtex PDF ArXiv
   \textcopyright 2017 IOP Publishing Ltd. One of the most challenging and recurring problems when modeling plasmas is the lack of data on the key atomic and molecular reactions that drive plasma processes. Even when there are data for some reactions, complete and validated datasets of chemistries are rarely available. This hinders research on plasma processes and curbs development of industrial applications. The QDB project aims to address this problem by providing a platform for provision, exchange, and validation of chemistry datasets. A new data model developed for QDB is presented. QDB collates published data on both electron scattering and heavy-particle reactions. These data are formed into reaction sets, which are then validated against experimental data where possible. This process produces both complete chemistry sets and identifies key reactions that are currently unreported in the literature. Gaps in the datasets can be filled using established theoretical methods. Initial validated chemistry sets for SF6/CF4/O2 and SF6/CF4/N2/H2 are presented as examples.

   @article{Tennyson2017, author = {Tennyson, J. and Rahimi, S. and Hill, C. and Tse, L. and Vibhakar, A. and Akello-Egwel, D. and Brown, D.B. and Dzarasova, A. and Hamilton, J.R. and Jaksch, D. and Mohr, S. and Wren-Little, K. and Bruckmeier, J. and Agarwal, A. and Bartschat, K. and Bogaerts, A. and Booth, J.-P. and Goeckner, M.J. and Hassouni, K. and Itikawa, Y. and Braams, B.J. and Krishnakumar, E. and Laricchiuta, A. and Mason, N.J. and Pandey, S. and Petrovic, Z.L. and Pu, Y.-K. and Ranjan, A. and Rauf, S. and Schulze, J. and Turner, M.M. and Ventzek, P. and Whitehead, J.C. and Yoon, J.-S.}, doi = {10.1088/1361-6595/aa6669}, issn = {13616595}, journal = {Plasma Sources Science and Technology}, keywords = {atomic and molecular data,database,plasma chemistry}, number = {5}, title = {"QDB: A new database of plasma chemistries and reactions"}, volume = {26}, year = {2017}, link = {https://iopscience.iop.org/article/10.1088/1361-6595/aa6669/pdf}, arxivid = {1704.04088} }

Posters

Online Matrix Completion with Side Information @ NeuRIPS 2020.

Miscellaneous.

Scientific communication

Articles written for Physics World:

• Superposition and entanglement flee the quantum nest. 21/06/2022.
• Quantum algorithm provides new approach to NP-hard problem. 01/06/2021.
• Quantum mechanics gives new insights into the Gibbs paradox. 20/03/2021.

Prizes

• IBM Qiskit Hackathon 2019 Europe: honourable mention for best game
• UCLQ Industry Day 2019: best research poster

Languages

Chinese (Mandarin), English, Dutch - Fluent

French - Advanced

Spanish - Intermediate

Chinese (Cantonese) - Intermediate listening, basic speaking

Virtual Bookshelf