I am a final-year undergraduate at the University of Cambridge, studying Computer Science.
My ambition is to develop powerful, yet safe and interpretable
abstract reasoners, whose
state and behaviour remain transparent to the end
To this end, I'm particularly interested in exploring how the mathematical toolkits we use to
understand and structure programs ‒ such as formal methods,
types and category theory ‒ can
inspire new ways
to both reverse-engineer existing neural networks, and build scalable neurosymbolic systems.
So far, my research has broadly focused on studying the behaviour of neural networks in
understanding both how they generalise when learning to perform abstract tasks, and
this tells us about the algorithms they've learned in order to do so.
I'm currently probing the foundations of neural
reasoning, exploring attacks on vision-language models, and poking language model
with a stick.
Image Hijacks: Adversarial Images can Control Generative Models at
Luke Bailey*, Euan Ong*, Stuart Russell, Scott Emmons (* denotes equal contribution)
project page + demo
We discovered that adversarial images can hijack the behaviour of vision-language models (VLMs) at
runtime. We developed a general method for crafting these image hijacks,
and trained image hijacks that force VLMs to output arbitrary text, leak their context window and
comply with harmful instructions.
Learnable Commutative Monoids for Graph Neural Networks
Euan Ong, Petar Veličković
Learning on Graphs, 2022
Using ideas from abstract algebra and functional programming, we built a new GNN aggregator that
beats the state of the art on complex aggregation problems (especially out-of-distribution) while
remaining efficient and parallelisable on large graphs.
Dissecting Deep Learning for Systematic Generalisation
Euan Ong, Etaash
Katiyar, Kai-En Chong, Albert
Informal research, 2021
We investigated the capabilities of transformers to systematically generalise when learning to
recognise formal languages (such as
Parity and 2-Dyck), empirically corroborating various theoretical claims about transformer generalisation.
by our observations, we
derived a parallel, stackless algorithm for recognising 2-Dyck that could (in principle) be
implemented by a transformer with a constant number of attention layers.
Object Detection in Thermal Imagery via Convolutional Neural
Euan Ong, Niki Trigoni, Pedro Porto Barque de Gusmão
Technical report, 2019
We trained a Faster R-CNN object detection network to identify landmarks (e.g. doors and windows) in
thermal images of indoor environments, with applications in the development of navigational aids for
search and rescue operations.