Matthew Jones

NeurIPS Conference 2025 Conference Paper

Bridging Equivariant GNNs and Spherical CNNs for Structured Physical Domains

• Colin Kohler
• Purvik Patel
• Nathan Vaska
• Justin Goodwin
• Matthew Jones
• Robert Platt
• Rajmonda Caceres
• Robin Walters

Many modeling tasks from disparate domains can be framed the same way, computing spherical signals from geometric inputs, for example, computing the radar response of different objects or navigating through an environment. This paper introduces G2Sphere, a general method for mapping object geometries to spherical signals. G2Sphere operates entirely in Fourier space, encoding geometric structure into latent Fourier features using equivariant neural networks and outputting the Fourier coefficients of the continuous target signal, which can be evaluated at any resolution. By utilizing a hybrid GNN-spherical CNN architecture, our method achieves much higher frequency output signal than comparable equivariant GNNs and avoids hand-engineered geometry features used previously by purely spherical methods. We perform experiments on various challenging domains including radar response modeling, aerodynamic drag prediction, and policy learning for manipulation and navigation. We find that G2Sphere outperforms competitive baselines in terms of accuracy and inference time, and we demonstrate that equivariance and Fourier features lead to improved sample efficiency and generalization. The source code is available at: https: //github. com/ColinKohler/geometry2sphere.

AAAI Conference 2023 Conference Paper

An Efficient Algorithm for Fair Multi-Agent Multi-Armed Bandit with Low Regret

• Matthew Jones
• Huy Nguyen
• Thy Nguyen

Recently a multi-agent variant of the classical multi-armed bandit was proposed to tackle fairness issues in online learning. Inspired by a long line of work in social choice and economics, the goal is to optimize the Nash social welfare instead of the total utility. Unfortunately previous algorithms either are not efficient or achieve sub-optimal regret in terms of the number of rounds. We propose a new efficient algorithm with lower regret than even previous inefficient ones. We also complement our efficient algorithm with an inefficient approach with regret that matches the lower bound for one agent. The experimental findings confirm the effectiveness of our efficient algorithm compared to the previous approaches.

AAMAS Conference 2023 Conference Paper

The Influence Maximisation Game

• Sukankana Chakraborty
• Sebastian Stein
• Ananthram Swami
• Matthew Jones
• Lewis Hill

The problem of influence maximisation investigates efficient ways in which external influence (typically limited by resources) can be applied to a social network to maximise control over the global behaviours of a population. It is an effective tool that finds its application in many real-world scenarios, for instance it can be used to gather intelligence in crowdsourcing activities and to incentivise people to adopt desirable public policies. While the problem has been studied extensively in theoretical settings, many of these approaches can be expensive and inefficient to apply in the real world, particularly when considering an unknown or irrational competitor. The influence maximisation game was designed to bridge this gap between theory and the practical application of this knowledge. In this experiment, human subjects are presented with networks where they can employ their own tactics to maintain maximum influence against a competitor (which in this case is an AI agent). We aim to determine how people strategise to spread influence in the real world. In particular, we determine if people always act rationally in these settings or if their strategies are inherently biased —in which case we aim to identify inexpensive, yet effective strategies that can outperform these biased strategies. Observing how people strategise in the real world can help us modify our theoretical results for more efficient practical applications.

AAAI Conference 2022 Conference Paper

Locally Private k-Means Clustering with Constant Multiplicative Approximation and Near-Optimal Additive Error

• Anamay Chaturvedi
• Matthew Jones
• Huy Lê Nguyễn

Given a data set of size n in d0 -dimensional Euclidean space, the k-means problem asks for a set of k points (called centers) such that the sum of the `2 2-distances between the data points and the set of centers is minimized. Previous work on this problem in the local differential privacy setting shows how to achieve multiplicative approximation factors arbitrarily close to optimal, but suffers high additive error. The additive error has also been seen to be an issue in implementations of differentially private k-means clustering algorithms in both the central and local settings. In this work, we introduce a new locally private k-means clustering algorithm that achieves near-optimal additive error whilst retaining constant multiplicative approximation factors and round complexity. Concretely, given any c > √ 2, our algorithm achieves O(k1+Õ(1/(c2 −2)) √ d0n log d0 poly log n) additive error with an O(c2 ) multiplicative approximation factor.

AAAI Conference 2021 Conference Paper

Differentially Private Clustering via Maximum Coverage

• Matthew Jones
• Huy L. Nguyen
• Thy D Nguyen

This paper studies the problem of clustering in metric spaces while preserving the privacy of individual data. Specifically, we examine differentially private variants of the k-medians and Euclidean k-means problems. We present polynomial algorithms with constant multiplicative error and lower additive error than the previous state-of-the-art for each problem. Additionally, our algorithms use a clustering algorithm without differential privacy as a black-box. This allows practitioners to control the trade-off between runtime and approximation factor by choosing a suitable clustering algorithm to use.

ICML Conference 2020 Conference Paper

Fair k-Centers via Maximum Matching

• Matthew Jones
• Huy L. Nguyen 0001
• Thy Dinh Nguyen

The field of algorithms has seen a push for fairness, or the removal of inherent bias, in recent history. In data summarization, where a much smaller subset of a data set is chosen to represent the whole of the data, fairness can be introduced by guaranteeing each "demographic group" a specific portion of the representative subset. Specifically, this paper examines this fair variant of the k-centers problem, where a subset of the data with cardinality k is chosen to minimize distance to the rest of the data. Previous papers working on this problem presented both a 3-approximation algorithm with a super-linear runtime and a linear-time algorithm whose approximation factor is exponential in the number of demographic groups. This paper combines the best of each algorithm by presenting a linear-time algorithm with a guaranteed 3-approximation factor and provides empirical evidence of both the algorithm’s runtime and effectiveness.

YNIMG Journal 2010 Journal Article

Serotonergic abnormalities in males with extreme levels of impulsive aggression, investigated using 11C-DASB and 11C-MDL100907 PET

• Angela J. Rylands
• Rainer Hinz
• Matthew Jones
• Elizabeth Barnett
• Michael E. Fairclough
• Steven Nolan
• Gavin Brown
• Adam W. McMahon

YNIMG Journal 2008 Journal Article

Dissociable roles of medial orbitofrontal cortex in human operant extinction learning

• Elizabeth C. Finger
• Derek G.V. Mitchell
• Matthew Jones
• R.J.R. Blair

Operant extinction, which features modification of instrumental responses to stimuli following a change in associated reinforcement, is an important form of learning for organisms in dynamic environments. Animal studies have highlighted orbital and medial prefrontal cortex and amygdala as mediators of operant extinction. Yet little is known about the neural mediators of operant extinction learning in humans. Using a novel fMRI paradigm, we report dissociable functional responses in distinct regions of medial orbitofrontal cortex (mOFC) during successful appetitive and aversive based operant extinction. During successful operant extinction, increased activity was observed in frontopolar OFC, while decreased activity was observed in caudal mOFC and rostral anterior cingulate cortex (rACC) relative to both (i) successful control trials where the reinforcement associated with the stimulus does not change; and (ii) successful acquisition trials during initial learning of the stimulus–reinforcement associations. Functional connectivity analysis demonstrated inverse connectivity between frontopolar OFC and both rACC and the amygdala. These data support animal models suggesting the importance of mOFC–amygdala interaction during operant extinction and expand our knowledge of the neural systems in humans. These findings suggest that in humans, frontopolar OFC modulates activity in caudal mOFC, rACC and amygdala during successful operant extinction learning.

YNIMG Journal 2007 Journal Article

Common regions of dorsal anterior cingulate and prefrontal–parietal cortices provide attentional control of distracters varying in emotionality and visibility

• Qian Luo
• Derek Mitchell
• Matthew Jones
• Krystal Mondillo
• Meena Vythilingam
• R. James R. Blair

Top–down attentional control is necessary to ensure successful task performance in the presence of distracters. Lateral prefrontal cortex, parietal cortex and anterior cingulate cortex have been previously implicated in top–down attentional control. However, it is unclear whether these regions are engaged independent of distracter type or whether, as has been suggested for anterior cingulate cortex, different regions provide attentional control over emotional versus other forms of salient distracter. In the current task, subjects viewed targets that were preceded by distracters that varied in both emotionality and visibility. We found that behaviorally, the presence of preceding distracters significantly interfered with target judgment. At the neural level, increases in the emotional and visual saliency of distracters were both associated with increased activity in proximal regions of prefrontal, parietal and cingulate cortex. Moreover, a conjunction analysis indicated considerable overlap in the regions of prefrontal, parietal cortex and anterior cingulate cortex responding to distracters of increased emotionality and visibility.

YNIMG Journal 2007 Journal Article

Neural dynamics for facial threat processing as revealed by gamma band synchronization using MEG

• Qian Luo
• Tom Holroyd
• Matthew Jones
• Talma Hendler
• James Blair

Facial threat conveys important information about imminent environmental danger. The rapid detection of this information is critical for survival and social interaction. However, due to technical and methodological difficulties, the spatiotemporal profile for facial threat processing is unknown. By utilizing magnetoencephalography (MEG), a brain-imaging technique with superb temporal resolution and fairly good spatial resolution, Synthetic Aperture Magnetometry (SAM), a recently developed source analysis technique, and a sliding window analysis, we identified the spatiotemporal development of facial threat processing in the gamma frequency band. We also tested the dual-route hypothesis by LeDoux who proposed, based on animal research, that there are two routes to the amygdala: a quick subcortical route and a slower and cortical route. Direct evidence with humans supporting this model has been lacking. Moreover, it has been unclear whether the subcortical route responds specifically to fearful expressions or to threatening expressions in general. We found early event-related synchronizations (ERS) in response to fearful faces in the hypothalamus/thalamus area (10–20 ms) and then the amygdala (20–30 ms). This was even earlier than the ERS response seen to fearful faces in visual cortex (40–50 ms). These data support LeDoux's suggestion of a quick, subcortical thamalo-amygdala route. Moreover, this route was specific for fear expressions; the ERS response in the amygdala to angry expressions had a late onset (150–160 ms). The ERS onset in prefrontal cortex followed that seen within the amygdala (around 160–210 ms). This is consistent with its role in higher-level emotional/cognitive processing.