Search

The FAIR Universe -- HiggsML Uncertainty Challenge focuses on measuring the physics properties of elementary particles with imperfect simulators due to differences in modelling systematic errors. Additionally, the challenge is leveraging a large-compute-scale AI platform for sharing datasets, training models, and hosting machine learning competitions. Our challenge brings together the physics and machine learning communities to advance our understanding and methodologies in handling systematic (epistemic) uncertainties within AI techniques., Comment: Whitepaper for the FAIR Universe HiggsML Uncertainty Challenge Competition, available : https://fair-universe.lbl.gov

Near horizon geometries of D$p$-branes with $p\neq 3$ are singular with a running dilaton. Bound states of D$p$ branes with their magnetic cousins, D$(6-p)$ branes, can stabilise the dilaton such that an AdS factor might appear in the near horizon region, potentially leading to a chain of AdS vacua of the form $AdS_{p+2}\times S^{p+2} \times \mathbb{T}^{6-2p}$. The solutions with $p=-1, 1, 3$ are supersymmetric with the cases $p=1, 3$ being well-known examples already. We construct the explicit (partially smeared) brane bound state solutions for all such configurations where the cases $p=-1,2$ are entirely novel and we find no AdS geometry for these. The two novel classes of solutions feature ghost branes (negative tension branes), and we suggest they are physical for the D$(-1)-$D$7$ solutions but unphysical for the D$2-$D$4$ solutions. The bound state of a D$(-1)$ and a D7 brane in supergravity was only hinted upon recently in \cite{Aguilar-Gutierrez:2022kvk} and here we correct the solution in order to preserve supersymmetry and find that the dilaton can indeed be stabilised which indicates there could be a holographic dual matrix theory which generalises the IKKT matrix model to allow for conformal invariance.

The recent development of Large Language Models (LLMs) has changed our role in interacting with them. Instead of primarily testing these models with questions we already know the answers to, we now use them to explore questions where the answers are unknown to us. This shift, which hasn't been fully addressed in existing datasets, highlights the growing need to understand naturally occurring human questions - that are more complex, open-ended, and reflective of real-world needs. To this end, we present NatQuest, a collection of 13,500 naturally occurring questions from three diverse sources: human-to-search-engine queries, human-to-human interactions, and human-to-LLM conversations. Our comprehensive collection enables a rich understanding of human curiosity across various domains and contexts. Our analysis reveals a significant presence of causal questions (up to 42%) within the dataset, for which we develop an iterative prompt improvement framework to identify all causal queries, and examine their unique linguistic properties, cognitive complexity, and source distribution. We also lay the groundwork to explore LLM performance on these questions and provide six efficient classification models to identify causal questions at scale for future work.

A topological interlocking assembly consists of rigid blocks together with a fixed frame, such that any subset of blocks is kinematically constrained and therefore cannot be removed from the assembly. In this paper we pursue a modular approach to construct (non-convex) interlocking blocks by combining finitely many tetrahedra and octahedra. This gives rise to polyhedra whose vertices can be described by the tetrahedral-octahedral honeycomb, also known as tetroctahedrille. We show that the resulting interlocking blocks are very versatile and allow many possibilities to form topological interlocking assemblies consisting of copies of a single block. We formulate a generalised construction of some of the introduced blocks to construct families of topological interlocking blocks. Moreover, we demonstrate a geometric application by using the tetroctahedrille to approximate given geometric objects. Finally, we show that given topological interlocking assemblies can be deformed continuously in order to obtain new topological interlocking assemblies., Comment: Keywords: Topological interlocking, Space Fillings, Triangulations, Origami, Approximations

Patents and scientific papers provide an essential source for measuring science and technology output, to be used as a basis for the most varied scientometric analyzes. Authors' and inventors' names are the key identifiers to carry out these analyses, which however, run up against the issue of disambiguation. By extension identifying inventors who are also academic authors is a non-trivial challenge. We propose a method using the International Patent Classification (IPC) and the IPCCAT API to assess the degree of similarity of patents and papers abstracts of a given inventor, in order to match both types of documents. The method is developed and manually qualified based on three corpora of patents extracted from the international EPO database Espacenet. Among a set of 4679 patents and 7720 inventors, we obtain 2501 authors. The proposed algorithm solves the general problem of disambiguation with an error rate lower than 5%., Comment: in French language

A significant challenge in multi-objective reinforcement learning is obtaining a Pareto front of policies that attain optimal performance under different preferences. We introduce Iterated Pareto Referent Optimisation (IPRO), a principled algorithm that decomposes the task of finding the Pareto front into a sequence of single-objective problems for which various solution methods exist. This enables us to establish convergence guarantees while providing an upper bound on the distance to undiscovered Pareto optimal solutions at each step. Empirical evaluations demonstrate that IPRO matches or outperforms methods that require additional domain knowledge. By leveraging problem-specific single-objective solvers, our approach also holds promise for applications beyond multi-objective reinforcement learning, such as in pathfinding and optimisation.

Frucht showed that, for any finite group $G$, there exists a cubic graph such that its automorphism group is isomorphic to $G$. For groups generated by two elements we simplify his construction to a graph with fewer nodes. In the general case, we address an oversight in Frucht's construction. We prove the existence of cycle double covers of the resulting graphs, leading to simplicial surfaces with given automorphism group. For almost all finite non-abelian simple groups we give alternative constructions based on graphic regular representations. In the general cases $C_n,D_n,A_5$ for $n\geq 4$, we provide alternative constructions of simplicial spheres. Furthermore, we embed these surfaces into the Euclidean 3-Space with equilateral triangles such that the automorphism group of the surface and the symmetry group of the corresponding polyhedron in $\mathrm{O}(3)$ are isomorphic.

The acquisition of survey responses is a crucial component in conducting research aimed at comprehending public opinion. However, survey data collection can be arduous, time-consuming, and expensive, with no assurance of an adequate response rate. In this paper, we propose a pioneering approach for predicting survey responses by examining quotations using machine learning. Our investigation focuses on evaluating the degree of favorability towards the United States, a topic of interest to many organizations and governments. We leverage a vast corpus of quotations from individuals across different nationalities and time periods to extract their level of favorability. We employ a combination of natural language processing techniques and machine learning algorithms to construct a predictive model for survey responses. We investigate two scenarios: first, when no surveys have been conducted in a country, and second when surveys have been conducted but in specific years and do not cover all the years. Our experimental results demonstrate that our proposed approach can predict survey responses with high accuracy. Furthermore, we provide an exhaustive analysis of the crucial features that contributed to the model's performance. This study has the potential to impact survey research in the field of data science by substantially decreasing the cost and time required to conduct surveys while simultaneously providing accurate predictions of public opinion., Comment: IEEE Swiss Conference on Data Science (SDS) 2023

In many risk-aware and multi-objective reinforcement learning settings, the utility of the user is derived from a single execution of a policy. In these settings, making decisions based on the average future returns is not suitable. For example, in a medical setting a patient may only have one opportunity to treat their illness. Making decisions using just the expected future returns -- known in reinforcement learning as the value -- cannot account for the potential range of adverse or positive outcomes a decision may have. Therefore, we should use the distribution over expected future returns differently to represent the critical information that the agent requires at decision time by taking both the future and accrued returns into consideration. In this paper, we propose two novel Monte Carlo tree search algorithms. Firstly, we present a Monte Carlo tree search algorithm that can compute policies for nonlinear utility functions (NLU-MCTS) by optimising the utility of the different possible returns attainable from individual policy executions, resulting in good policies for both risk-aware and multi-objective settings. Secondly, we propose a distributional Monte Carlo tree search algorithm (DMCTS) which extends NLU-MCTS. DMCTS computes an approximate posterior distribution over the utility of the returns, and utilises Thompson sampling during planning to compute policies in risk-aware and multi-objective settings. Both algorithms outperform the state-of-the-art in multi-objective reinforcement learning for the expected utility of the returns., Comment: arXiv admin note: substantial text overlap with arXiv:2102.00966

The discovery of neutrino oscillation, proving that neutrinos do have masses, reveals the misfits of particles in the current Standard Model (SM) theory. In theory, neutrinos having masses could result in lepton flavour not being a symmetry called Lepton Flavour Violation (LFV). While SM theory extensions allowed LFV processes, their branching fractions are too small, making them unobservable even with the strongest equipment up-to-date. With that, scientists in recent years have generated LFV-like processes from the combined LHCb and Monte-Carlo-Simulated data in an attempt to identify LFV using Artificial Intelligence (AI), specifically Machine Learning (ML) and Deep Learning (DL). In this paper, the performance of several algorithms in AI has been presented, such as XGBoost, LightGBM, custom 1-D Dense Block Neural Networks (DBNNs), and custom 1-D Convolutional Neural Networks (CNNs) in identifying LFV signals, specifically $\tau^{-} \rightarrow \mu^{-}\mu^{-}\mu^{+}$ decay from the combined LHCb and Monte-Carlo-Simulated data that imitates the signatures of the said decay. Kolmogorov-Smirnov (KS) and Cramer-von Mises (CvM) tests were also conducted to verify the validity of predictions for each of the trained algorithms. The result shows decent performances among algorithms, except for the LightGBM, for failing the CvM test, and a 20-layered CNN for having recorded a considerably low AUC. Meanwhile, XGBoost and a 10-layered DBNN recorded the highest AUC of 0.88. The main contribution of this paper is the extensive experiment involving custom DBNN and CNN algorithms in different layers, all of which have been rarely used in the past years in identifying LFV-like signatures, unlike GBMs and tree-based algorithms, which have been more popular in the said task., Comment: 6 pages, 5 figures, 3 tables

We present a description of the GBAR positron (e+) trapping apparatus, which consists of a three stage Buffer Gas Trap (BGT) followed by a High Field Penning Trap (HFT), and discuss its performance. The overall goal of the GBAR experiment is to measure the acceleration of the neutral antihydrogen (H) atom in the terrestrial gravitational field by neutralising a positive antihydrogen ion (H+), which has been cooled to a low temperature, and observing the subsequent H annihilation following free fall. To produce one H+ ion, about 10^10 positrons, efficiently converted into positronium (Ps), together with about 10^7 antiprotons (p), are required. The positrons, produced from an electron linac-based system, are accumulated first in the BGT whereafter they are stacked in the ultra-high vacuum HFT, where we have been able to trap 1.4(2) x 10^9 positrons in 1100 seconds.

In multi-objective optimization, learning all the policies that reach Pareto-efficient solutions is an expensive process. The set of optimal policies can grow exponentially with the number of objectives, and recovering all solutions requires an exhaustive exploration of the entire state space. We propose Pareto Conditioned Networks (PCN), a method that uses a single neural network to encompass all non-dominated policies. PCN associates every past transition with its episode's return. It trains the network such that, when conditioned on this same return, it should reenact said transition. In doing so we transform the optimization problem into a classification problem. We recover a concrete policy by conditioning the network on the desired Pareto-efficient solution. Our method is stable as it learns in a supervised fashion, thus avoiding moving target issues. Moreover, by using a single network, PCN scales efficiently with the number of objectives. Finally, it makes minimal assumptions on the shape of the Pareto front, which makes it suitable to a wider range of problems than previous state-of-the-art multi-objective reinforcement learning algorithms., Comment: Accepted at the International Conference on Autonomous Agents and Multiagent Systems (AAMAS) 2022

Infectious disease outbreaks can have a disruptive impact on public health and societal processes. As decision making in the context of epidemic mitigation is hard, reinforcement learning provides a methodology to automatically learn prevention strategies in combination with complex epidemic models. Current research focuses on optimizing policies w.r.t. a single objective, such as the pathogen's attack rate. However, as the mitigation of epidemics involves distinct, and possibly conflicting criteria (i.a., prevalence, mortality, morbidity, cost), a multi-objective approach is warranted to learn balanced policies. To lift this decision-making process to real-world epidemic models, we apply deep multi-objective reinforcement learning and build upon a state-of-the-art algorithm, Pareto Conditioned Networks (PCN), to learn a set of solutions that approximates the Pareto front of the decision problem. We consider the first wave of the Belgian COVID-19 epidemic, which was mitigated by a lockdown, and study different deconfinement strategies, aiming to minimize both COVID-19 cases (i.e., infections and hospitalizations) and the societal burden that is induced by the applied mitigation measures. We contribute a multi-objective Markov decision process that encapsulates the stochastic compartment model that was used to inform policy makers during the COVID-19 epidemic. As these social mitigation measures are implemented in a continuous action space that modulates the contact matrix of the age-structured epidemic model, we extend PCN to this setting. We evaluate the solution returned by PCN, and observe that it correctly learns to reduce the social burden whenever the hospitalization rates are sufficiently low. In this work, we thus show that multi-objective reinforcement learning is attainable in complex epidemiological models and provides essential insights to balance complex mitigation policies.

Assuming that the anthropogenic impact of visitors to a natural park can be reduced by communication actions, we have drawn up a bioacoustical protocol combined with a protocol to measure the effectiveness of the communication actions. A simple eco-awareness communication method will be compared to a binding communication method. We address the questions: does visitors' eco-awareness drive them to change their attitudes and behavior in order to limit their own noise disturbance, and does commitment improve their awareness? We detail here the main concepts considered and the methodology followed to discuss the preliminary results.

Many recent successful off-policy multi-agent reinforcement learning (MARL) algorithms for cooperative partially observable environments focus on finding factorized value functions, leading to convoluted network structures. Building on the structure of independent Q-learners, our LAN algorithm takes a radically different approach, leveraging a dueling architecture to learn for each agent a decentralized best-response policies via individual advantage functions. The learning is stabilized by a centralized critic whose primary objective is to reduce the moving target problem of the individual advantages. The critic, whose network's size is independent of the number of agents, is cast aside after learning. Evaluation on the StarCraft II multi-agent challenge benchmark shows that LAN reaches state-of-the-art performance and is highly scalable with respect to the number of agents, opening up a promising alternative direction for MARL research., Comment: https://openreview.net/forum?id=adpKzWQunW