Publications

AfriHate: A Multilingual Collection of Hate Speech and Abusive Language Datasets for African Languages
Shamsuddeen Hassan Muhammad
Idris Abdulmumin
Abinew Ayele
David Ifeoluwa Adelani
Ibrahim Ahmad
Saminu Mohammad Aliyu
Nelson Odhiambo Onyango
Lilian D. A. Wanzare
Samuel Rutunda
Lukman Jibril Aliyu
Esubalew Alemneh
Oumaima Hourrane
Hagos Gebremichael
Elyas Abdi Ismail
Meriem Beloucif
Ebrahim Chekol Jibril
Andiswa Bukula
Rooweither Mabuya
Salomey Osei
Abigail Oppong … (see 7 more)
Tadesse Belay
Tadesse Kebede Guge
Tesfa Tegegne Asfaw
Chiamaka Ijeoma Chukwuneke
Paul Rottger
Seid Muhie Yimam
Nedjma OUSIDHOUM
Hate speech and abusive language are global phenomena that need socio-cultural background knowledge to be understood, identified, and modera… (see more)ted. However, in many regions of the Global South, there have been several documented occurrences of (1) absence of moderation and (2) censorship due to the reliance on keyword spotting out of context. Further, high-profile individuals have frequently been at the center of the moderation process, while large and targeted hate speech campaigns against minorities have been overlooked. These limitations are mainly due to the lack of high-quality data in the local languages and the failure to include local communities in the collection, annotation, and moderation processes. To address this issue, we present AfriHate: a multilingual collection of hate speech and abusive language datasets in 15 African languages. Each instance in AfriHate is annotated by native speakers familiar with the local culture. We report the challenges related to the construction of the datasets and present various classification baseline results with and without using LLMs. The datasets, individual annotations, and hate speech and offensive language lexicons are available on https://github.com/AfriHate/AfriHate
2025-01-13
ArXiv (preprint)
doi.org
arxiv.org
Integrating food webs in species distribution models can improve ecological niche estimation and predictions
Giovanni Poggiato
Jérémy Andréoletti
Laura J. Pollock
Wilfried Thuiller
2025-01-13
Ecography (published)
doi.org
Towards Robust Saliency Maps
Nham Le
Arie Gurfinkel
Xujie Si
Chuqin Geng
Saliency maps are one of the most popular tools to interpret the operation of a neural network: they compute input features deemed relevant … (see more)to the final prediction, which are often subsets of pixels that are easily understandable by a human being. However, it is known that relying solely on human assessment to judge a saliency map method can be misleading. In this work, we propose a new neural network verification specification called saliency-robustness, which aims to use formal methods to prove a relationship between Vanilla Gradient (VG) -- a simple yet surprisingly effective saliency map method -- and the network's prediction: given a network, if an input
2025-01-13
Proceedings of the 16th Asian Conference on Machine Learning (published)
proceedings.mlr.press
proceedings.mlr.press
Multi-agent deep reinforcement learning with online and fair optimal dispatch of EV aggregators
Arian Shah Kamrani
Anoosh Dini
Hanane Dagdougui
Keyhan Sheshyekani
The growing popularity of electric vehicles (EVs) and the unpredictable behavior of EV owners have attracted attention to real-time coordina… (see more)tion of EVs charging management. This paper presents a hierarchical structure for charging management of EVs by integrating fairness and efficiency concepts within the operations of the distribution system operator (DSO) while utilizing a multi-agent deep reinforcement learning (MADRL) framework to tackle the complexities of energy purchasing and distribution among EV aggregators (EVAs). At the upper level, DSO calculates the maximum allowable power for each EVA based on power flow constraints to ensure grid safety. Then, it finds the optimal efficiency-jain tradeoff (EJT) point, where it sells the highest energy amount while ensuring equitable energy distribution. At the lower level, initially, each EVA acts as an agent employing a double deep Q-network (DDQN) with adaptive learning rates and prioritized experience replay to determine optimal energy purchases from the DSO. Then, the real-time smart dispatch (RSD) controller prioritizes EVs for energy dispatch based on relevant EVs information. Findings indicate the proposed enhanced DDQN outperforms deep deterministic policy gradient (DDPG) and proximal policy optimization (PPO) in cumulative rewards and convergence speed. Finally, the framework’s performance is evaluated against uncontrolled charging and the first come first serve (FCFS) scenario using the 118-bus distribution system, demonstrating superior performance in maintaining safe operation of the grid while reducing charging costs for EVAs. Additionally, the framework’s integration with renewable energy sources (RESs), such as photovoltaic (PV), demonstrates its potential to enhance grid reliability. • Introduces a scalable MADRL framework for real-time EV charging and energy distribution. • Ensures fairness via an Efficiency-Jain Tradeoff (EJT) strategy at the DSO level. • Enhances agent convergence with DDQN using adaptive learning rates and prioritized replay. • Preserves stakeholder privacy with decentralized control and minimal data sharing. • Balances grid reliability with equitable energy allocation under dynamic uncertainties.
2025-01-08
Machine Learning with Applications (published)
doi.org
Open Problems in Machine Unlearning for AI Safety
Fazl Barez
Tingchen Fu
Ameya Prabhu
Stephen Casper
Amartya Sanyal
Adel Bibi
Aidan O'Gara
Robert Kirk
Benjamin Bucknall
Timothy Fist
Luke Ong
Philip Torr
Kwok-Yan Lam
Robert Trager
David M. Krueger
Sören Mindermann
Jose Hernandez-Orallo
Mor Geva
Yarin Gal
As AI systems become more capable, widely deployed, and increasingly autonomous in critical areas such as cybersecurity, biological research… (see more), and healthcare, ensuring their safety and alignment with human values is paramount. Machine unlearning -- the ability to selectively forget or suppress specific types of knowledge -- has shown promise for privacy and data removal tasks, which has been the primary focus of existing research. More recently, its potential application to AI safety has gained attention. In this paper, we identify key limitations that prevent unlearning from serving as a comprehensive solution for AI safety, particularly in managing dual-use knowledge in sensitive domains like cybersecurity and chemical, biological, radiological, and nuclear (CBRN) safety. In these contexts, information can be both beneficial and harmful, and models may combine seemingly harmless information for harmful purposes -- unlearning this information could strongly affect beneficial uses. We provide an overview of inherent constraints and open problems, including the broader side effects of unlearning dangerous knowledge, as well as previously unexplored tensions between unlearning and existing safety mechanisms. Finally, we investigate challenges related to evaluation, robustness, and the preservation of safety features during unlearning. By mapping these limitations and open challenges, we aim to guide future research toward realistic applications of unlearning within a broader AI safety framework, acknowledging its limitations and highlighting areas where alternative approaches may be required.
2025-01-08
ArXiv (preprint)
doi.org
arxiv.org
Soup to go: mitigating forgetting during continual learning with model averaging
Anat Kleiman
Gintare Karolina Dziugaite
Jonathan Frankle
Sham M. Kakade
Mansheej Paul
In continual learning, where task data arrives in a sequence, fine-tuning on later tasks will often lead to performance degradation on earli… (see more)er tasks. This is especially pronounced when these tasks come from diverse domains. In this setting, how can we mitigate catastrophic forgetting of earlier tasks and retain what the model has learned with minimal computational expenses? Inspired by other merging methods, and L2-regression, we propose Sequential Fine-tuning with Averaging (SFA), a method that merges currently training models with earlier checkpoints during the course of training. SOTA approaches typically maintain a data buffer of past tasks or impose a penalty at each gradient step. In contrast, our method achieves comparable results without the need to store past data, or multiple copies of parameters for each gradient step. Furthermore, our method outperforms common merging techniques such as Task Arithmetic, TIES Merging, and WiSE-FT, as well as other penalty methods like L2 and Elastic Weight Consolidation. In turn, our method offers insight into the benefits of merging partially-trained models during training across both image and language domains.
2025-01-08
ArXiv (preprint)
doi.org
openreview.net
The Harmonic Exponential Filter for Nonparametric Estimation on Motion Groups
Miguel Saavedra-Ruiz
Steven A. Parkison
Ria Arora
James Richard Forbes
Liam Paull
Bayesian estimation is a vital tool in robotics as it allows systems to update the robot state belief using incomplete information from nois… (see more)y sensors. To render the state estimation problem tractable, many systems assume that the motion and measurement noise, as well as the state distribution, are unimodal and Gaussian. However, there are numerous scenarios and systems that do not comply with these assumptions. Existing nonparametric filters that are used to model multimodal distributions have drawbacks that limit their ability to represent a diverse set of distributions. This paper introduces a novel approach to nonparametric Bayesian filtering on motion groups, designed to handle multimodal distributions using harmonic exponential distributions. This approach leverages two key insights of harmonic exponential distributions: a) the product of two distributions can be expressed as the element-wise addition of their log-likelihood Fourier coefficients, and b) the convolution of two distributions can be efficiently computed as the tensor product of their Fourier coefficients. These observations enable the development of an efficient and asymptotically exact solution to the Bayes filter up to the band limit of a Fourier transform. We demonstrate our filter's performance compared with established nonparametric filtering methods across simulated and real-world localization tasks.
2025-01-07
IEEE Robotics and Automation Letters (published)
doi.org
arxiv.org
Adaptive Experiments Under Data Sparse Settings: Applications for Educational Platforms
Haochen Song
Ilya Musabirov
Ananya Bhattacharjee
Audrey Durand
Meredith Franklin
Anna Rafferty
Joseph Jay Williams
Adaptive experimentation is increasingly used in educational platforms to personalize learning through dynamic content and feedback. However… (see more), standard adaptive strategies such as Thompson Sampling often underperform in real-world educational settings where content variations are numerous and student participation is limited, resulting in sparse data. In particular, Thompson Sampling can lead to imbalanced content allocation and delayed convergence on which aspects of content are most effective for student learning. To address these challenges, we introduce Weighted Allocation Probability Adjusted Thompson Sampling (WAPTS), an algorithm that refines the sampling strategy to improve content-related decision-making in data-sparse environments. WAPTS is guided by the principle of lenient regret, allowing near-optimal allocations to accelerate learning while still exploring promising content. We evaluate WAPTS in a learnersourcing scenario where students rate peer-generated learning materials, and demonstrate that it enables earlier and more reliable identification of promising treatments.
2025-01-06
ArXiv (preprint)
arxiv.org
arxiv.org
L’appréhension empirique du leadership normatif d’une organisation internationale : l’exemple de l’Organisation mondiale de la Santé
Catherine Régis
Gaëlle Foucault
Pierre Larouche
Jean-Louis Denis
Miriam Cohen
En plein essor, la recherche empirique en droit participe à la création de nouvelles connaissances et ouvre aux juristes d’autres voies … (see more)pour étudier une question, un phénomène. Oser l’empirisme n’est pas chose aisée, mais les auteurs du présent article ont pris ce virage et proposent d’en exposer le récit. En construisant deux méthodes distinctes (pour deux projets), ils ont pu tester les possibilités qu’offre la recherche empirique pour appréhender l’enjeu du leadership normatif de l’Organisation mondiale de la Santé (OMS). Destiné à aiguiller à partir d’une expérience celles et ceux qui voudraient s’aventurer dans l’empirisme, cet article met en lumière les défis rencontrés, mais surtout les atouts d’une telle recherche. La richesse des informations obtenues a en effet grandement bonifié la compréhension de la trajectoire des normes de l’OMS et de leurs impacts sur les États.
2025-01-06
Lex Electronica (published)
doi.org
The empirical apprehension of the normative leadership of an international organization: The example of the World Health Organization
Catherine Régis
Gaëlle Foucault
Pierre Larouche
Jean-Louis Denis
Miriam Cohen
2025-01-06
Lex Electronica (published)
doi.org
A video-based approach to decipher intubation decisions for the critically ill
Guillaume Dumas
Jean-Rémi Lavillegrand
Elie Azoulay
2025-01-06
Intensive Care Medicine (published)
doi.org
IRIS: A Bayesian Approach for Image Reconstruction in Radio Interferometry with expressive Score-Based priors
No'e Dia
M. J. Yantovski-Barth
Alexandre Adam
Micah Bowles
Laurence Perreault-Levasseur
Yashar Hezaveh
Anna M. M. Scaife
Inferring sky surface brightness distributions from noisy interferometric data in a principled statistical framework has been a key challeng… (see more)e in radio astronomy. In this work, we introduce Imaging for Radio Interferometry with Score-based models (IRIS). We use score-based models trained on optical images of galaxies as an expressive prior in combination with a Gaussian likelihood in the uv-space to infer images of protoplanetary disks from visibility data of the DSHARP survey conducted by ALMA. We demonstrate the advantages of this framework compared with traditional radio interferometry imaging algorithms, showing that it produces plausible posterior samples despite the use of a misspecified galaxy prior. Through coverage testing on simulations, we empirically evaluate the accuracy of this approach to generate calibrated posterior samples.
2025-01-04
ArXiv (preprint)
doi.org
arxiv.org