Tootfinder

HOLA: Enhancing Audio-visual Deepfake Detection via Hierarchical Contextual Aggregations and Efficient Pre-training
Xuecheng Wu, Danlei Huang, Heli Sun, Xinyi Yin, Yifan Wang, Hao Wang, Jia Zhang, Fei Wang, Peihao Guo, Suyu Xing, Junxiao Xue, Liang He
https://arxiv.org/abs/2507.22781

HOLA: Enhancing Audio-visual Deepfake Detection via Hierarchical Contextual Aggregations and Efficient Pre-training
Advances in Generative AI have made video-level deepfake detection increasingly challenging, exposing the limitations of current detection techniques. In this paper, we present HOLA, our solution to the Video-Level Deepfake Detection track of 2025 1M-Deepfakes Detection Challenge. Inspired by the success of large-scale pre-training in the general domain, we first scale audio-visual self-supervised pre-training in the multimodal video-level deepfake detection, which leverages our self-built data…

Stabilizing Humanoid Robot Trajectory Generation via Physics-Informed Learning and Control-Informed Steering
Evelyn D'Elia, Paolo Maria Viceconte, Lorenzo Rapetti, Diego Ferigo, Giulio Romualdi, Giuseppe L'Erario, Raffaello Camoriano, Daniele Pucci
https://arxiv.org/abs/2509.24697

Stabilizing Humanoid Robot Trajectory Generation via Physics-Informed Learning and Control-Informed Steering
Recent trends in humanoid robot control have successfully employed imitation learning to enable the learned generation of smooth, human-like trajectories from human data. While these approaches make more realistic motions possible, they are limited by the amount of available motion data, and do not incorporate prior knowledge about the physical laws governing the system and its interactions with the environment. Thus they may violate such laws, leading to divergent trajectories and sliding cont…

A modular framework for collaborative human-AI, multi-modal and multi-beamline synchrotron experiments
Adam A. Corrao, Phillip M. Maffettone, Bruce Ravel, Thomas A. Caswell, Howie Joress, Stuart I. Wilkins, Daniel Olds
https://arxiv.org/abs/2509.22959

A modular framework for collaborative human-AI, multi-modal and multi-beamline synchrotron experiments
High-throughput materials discovery and studies of complex functional materials increasingly rely on multi-modal characterization performed at synchrotron light sources. However, measurements are typically done with no use of data until after an experiment, neglecting opportunities for data-driven insights to guide measurements. We developed a modular, open-source framework that incorporates artificial intelligence within the Bluesky control and data streaming infrastructure at NSLS-II, enablin…

TL;DR: what if instead of denying the harms of fascism, we denied its suppressive threats of punishment
Many of us have really sharpened our denial skills since the advent of the ongoing pandemic (perhaps you even hesitated at the word "ongoing" there and thought "maybe I won't read this one, it seems like it'll be tiresome"). I don't say this as a preface to a fiery condemnation or a plea to "sanity" or a bunch of evidence of how bad things are, because I too have honed my denial skills in these recent years, and I feel like talking about that development.
Denial comes in many forms, including strategic information avoidance ("I don't have time to look that up right now", "I keep forgetting to look into that", "well this author made a tiny mistake, so I'll click away and read something else", "I'm so tired of hearing about this, let me scroll farther", etc.) strategic dismissal ("look, there's a bit of uncertainty here, I should ignore this", "this doesn't line up perfectly with my anecdotal experience, it must be completely wrong", etc.) and strategic forgetting ("I don't remember what that one study said exactly; it was painful to think about", "I forgot exactly what my friend was saying when we got into that argument", etc.). It's in fact a kind of skill that you can get better at, along with the complementary skill of compartmentalization. It can of course be incredibly harmful, and a huge genre of fables exists precisely to highlight its harms, but it also has some short-term psychological benefits, chiefly in the form of muting anxiety. This is not an endorsement of denial (the harms can be catastrophic), but I want to acknowledge that there *are* short-term benefits. Via compartmentalization, it's even possible to be honest with ourselves about some of our own denials without giving them up immediately.
But as I said earlier, I'm not here to talk you out of your denials. Instead, given that we are so good at denial now, I'm here to ask you to be strategic about it. In particular, we live in a world awash with propaganda/advertising that serves both political and commercial ends. Why not use some of our denial skills to counteract that?
For example, I know quite a few people in complete denial of our current political situation, but those who aren't (including myself) often express consternation about just how many people in the country are supporting literal fascism. Of course, logically that appearance of widespread support is going to be partly a lie, given how much our public media is beholden to the fascists or outright in their side. Finding better facts on the true level of support is hard, but in the meantime, why not be in denial about the "fact" that Trump has widespread popular support?
To give another example: advertisers constantly barrage us with messages about our bodies and weight, trying to keep us insecure (and thus in the mood to spend money to "fix" the problem). For sure cutting through that bullshit by reading about body positivity etc. is a better solution, but in the meantime, why not be in denial about there being anything wrong with your body?
This kind of intentional denial certainly has its own risks (our bodies do actually need regular maintenance, for example, so complete denial on that front is risky) but there's definitely a whole lot of misinformation out there that it would be better to ignore. To the extent such denial expands to a more general denial of underlying problems, this idea of intentional denial is probably just bad. But I sure wish that in a world where people (including myself) routinely deny significant widespread dangers like COVID-19's long-term risks or the ongoing harms of escalating fascism, they'd at least also deny some of the propaganda keeping them unhappy and passive. Instead of being in denial about US-run concentration camps, why not be in denial that the state will be able to punish you for resisting them?

Caught in the Act: a mechanistic approach to detecting deception
Gerard Boxo, Ryan Socha, Daniel Yoo, Shivam Raval
https://arxiv.org/abs/2508.19505 https://

Caught in the Act: a mechanistic approach to detecting deception
Sophisticated instrumentation for AI systems might have indicators that signal misalignment from human values, not unlike a "check engine" light in cars. One such indicator of misalignment is deceptiveness in generated responses. Future AI instrumentation may have the ability to detect when an LLM generates deceptive responses while reasoning about seemingly plausible but incorrect answers to factual questions. In this work, we demonstrate that linear probes on LLMs internal activations can det…

Optimizing Inference in Transformer-Based Models: A Multi-Method Benchmark
Siu Hang Ho, Prasad Ganesan, Nguyen Duong, Daniel Schlabig
https://arxiv.org/abs/2509.17894 https://…

Optimizing Inference in Transformer-Based Models: A Multi-Method Benchmark
Efficient inference is a critical challenge in deep generative modeling, particularly as diffusion models grow in capacity and complexity. While increased complexity often improves accuracy, it raises compute costs, latency, and memory requirements. This work investigates techniques such as pruning, quantization, knowledge distillation, and simplified attention to reduce computational overhead without impacting performance. The study also explores the Mixture of Experts (MoE) approach to furthe…

SuryaBench: Benchmark Dataset for Advancing Machine Learning in Heliophysics and Space Weather Prediction
Sujit Roy, Dinesha V. Hegde, Johannes Schmude, Amy Lin, Vishal Gaur, Rohit Lal, Kshitiz Mandal, Talwinder Singh, Andr\'es Mu\~noz-Jaramillo, Kang Yang, Chetraj Pandey, Jinsu Hong, Berkay Aydin, Ryan McGranaghan, Spiridon Kasapis, Vishal Upendran, Shah Bahauddin, Daniel da Silva, Marcus Freitag, Iksha Gurung, Nikolai Pogorelov, Campbell Watson, Manil Maskey, Juan Bernabe-Moreno,…

SuryaBench: Benchmark Dataset for Advancing Machine Learning in Heliophysics and Space Weather Prediction
This paper introduces a high resolution, machine learning-ready heliophysics dataset derived from NASA's Solar Dynamics Observatory (SDO), specifically designed to advance machine learning (ML) applications in solar physics and space weather forecasting. The dataset includes processed imagery from the Atmospheric Imaging Assembly (AIA) and Helioseismic and Magnetic Imager (HMI), spanning a solar cycle from May 2010 to July 2024. To ensure suitability for ML tasks, the data has been preprocessed…

Super-Resolution of Sentinel-2 Images Using a Geometry-Guided Back-Projection Network with Self-Attention
Ivan Pereira-S\'anchez, Daniel Torres, Francesc Alcover, Bartomeu Garau, Julia Navarro, Catalina Sbert, Joan Duran
https://arxiv.org/abs/2508.04729

Super-Resolution of Sentinel-2 Images Using a Geometry-Guided Back-Projection Network with Self-Attention
The Sentinel-2 mission provides multispectral imagery with 13 bands at resolutions of 10m, 20m, and 60m. In particular, the 10m bands offer fine structural detail, while the 20m bands capture richer spectral information. In this paper, we propose a geometry-guided super-resolution model for fusing the 10m and 20m bands. Our approach introduces a cluster-based learning procedure to generate a geometry-rich guiding image from the 10m bands. This image is integrated into an unfolded back-projectio…

4D-QENS Analysis of Correlated Ionic Conduction in SrCl$_2$
Jared Coles, Omar Chmaissem, Matthew Krogstad, Daniel M. Pajerowski, Feng Ye, Duck Young Chung, Mercouri G. Kanatzidis, Stephan Rosenkranz, Raymond Osborn
https://arxiv.org/abs/2509.20312

4D-QENS Analysis of Correlated Ionic Conduction in SrCl$_2$
Methods of elucidating the mechanisms of fast-ion conduction in solid-state materials are pivotal for advancements in energy technologies such as batteries, fuel cells, sensors, and supercapacitors. In this study, we examine the ionic conduction pathways in single crystal SrCl$_2$, which is a fast-ion conductor above 900~K, using four-dimensional Quasi-Elastic Neutron Scattering (4D-QENS). We explore both coherent and incoherent neutron scattering at temperatures above the transition temperatur…

Evaluation Guidelines for Empirical Studies in Software Engineering involving LLMs
Sebastian Baltes, Florian Angermeir, Chetan Arora, Marvin Mu\~noz Bar\'on, Chunyang Chen, Lukas B\"ohme, Fabio Calefato, Neil Ernst, Davide Falessi, Brian Fitzgerald, Davide Fucci, Marcos Kalinowski, Stefano Lambiase, Daniel Russo, Mircea Lungu, Lutz Prechelt, Paul Ralph, Christoph Treude, Stefan Wagner

Evaluation Guidelines for Empirical Studies in Software Engineering involving LLMs
Large language models (LLMs) are increasingly being integrated into software engineering (SE) research and practice, yet their non-determinism, opaque training data, and evolving architectures complicate the reproduction and replication of empirical studies. We present a community effort to scope this space, introducing a taxonomy of LLM-based study types together with eight guidelines for designing and reporting empirical studies involving LLMs. The guidelines present essential (must) criteria…