Tootfinder

ADMIRE-BayesOpt: Accelerated Data MIxture RE-weighting for Language Models with Bayesian Optimization
Shengzhuang Chen, Xu Ouyang, Michael Arthur Leopold Pearce, Thomas Hartvigsen, Jonathan Richard Schwarz
https://arxiv.org/abs/2508.11551

ADMIRE-BayesOpt: Accelerated Data MIxture RE-weighting for Language Models with Bayesian Optimization
Determining the optimal data mixture for large language model training remains a challenging problem with an outsized impact on performance. In practice, language model developers continue to rely on heuristic exploration since no learning-based approach has emerged as a reliable solution. In this work, we propose to view the selection of training data mixtures as a black-box hyperparameter optimization problem, for which Bayesian Optimization is a well-established class of appropriate algorith…

Undersampled Phase Retrieval with Image Priors
Stanislas Ducotterd, Zhiyuan Hu, Michael Unser, Jonathan Dong
https://arxiv.org/abs/2509.15026 https://arxiv…

Undersampled Phase Retrieval with Image Priors
Phase retrieval seeks to recover a complex signal from amplitude-only measurements, a challenging nonlinear inverse problem. Current theory and algorithms often ignore signal priors. By contrast, we evaluate here a variety of image priors in the context of severe undersampling with structured random Fourier measurements. Our results show that those priors significantly improve reconstruction, allowing accurate reconstruction even below the weak recovery threshold.

Scalable, Technology-Agnostic Diagnosis and Predictive Maintenance for Point Machine using Deep Learning
Eduardo Di Santi (Digital and Integrated Systems, Alstom), Ruixiang Ci (Innovation and Smart Mobility, Alstom), Cl\'ement Lefebvre (Digital and Integrated Systems, Alstom), Nenad Mijatovic (Digital and Integrated Systems, Alstom), Michele Pugnaloni (Digital and Integrated Systems, Alstom), Jonathan Brown (Digital and Integrated Systems, Alstom), Victor Mart\'in (Digital and …

Scalable, Technology-Agnostic Diagnosis and Predictive Maintenance for Point Machine using Deep Learning
The Point Machine (PM) is a critical piece of railway equipment that switches train routes by diverting tracks through a switchblade. As with any critical safety equipment, a failure will halt operations leading to service disruptions; therefore, pre-emptive maintenance may avoid unnecessary interruptions by detecting anomalies before they become failures. Previous work relies on several inputs and crafting custom features by segmenting the signal. This not only adds additional requirements for…

An Improved Solution to the Two Normal Means Problem via Regularization
Yang Liu, Jonathan P. Williams
https://arxiv.org/abs/2508.13012 https://arxiv.org/p…

An Improved Solution to the Two Normal Means Problem via Regularization
The many-normal-means problem is a classic example that motivates the development of many important inferential procedures in the history of statistics. In this short note, we consider a further special case of the problem, which involves only two normally distributed data points with a constraint that the pair of means are not too far apart from one another. Starting with a regularized ML estimator, we construct a novel possibilistic IM for marginal inference on one of the two means. Not only …

Homological torsion growth in non-normal chains of graphs of free groups
Dario Ascari, Jonathan Fruchter
https://arxiv.org/abs/2509.15075 https://arxiv.org…

Homological torsion growth in non-normal chains of graphs of free groups
Let $G$ be a hyperbolic group that splits as a graph of free groups with cyclic edge groups, and which is not isomorphic to a free product of free and surface groups. We show that $G$ admits an exhausting, nested sequence of finite-index non-normal subgroups $G\ge G_1 \ge G_2 \ge \cdots$ with exponential homological torsion growth. More specifically, we prove that simultaneously for every prime $p$, $\liminf_{n\rightarrow \infty} \frac{\log \vert \mathrm{Tor}_p(G_n^{\mathrm{ab}})\vert}{[G:G_n]}…

From Misunderstandings to Learning Opportunities: Leveraging Generative AI in Discussion Forums to Support Student Learning
Stanislav Pozdniakov, Jonathan Brazil, Oleksandra Poquet, Stephan Krusche, Santiago Berrezueta-Guzman, Shazia Sadiq, Hassan Khosravi
https://arxiv.org/abs/2508.11150

From Misunderstandings to Learning Opportunities: Leveraging Generative AI in Discussion Forums to Support Student Learning
In the contemporary educational landscape, particularly in large classroom settings, discussion forums have become a crucial tool for promoting interaction and addressing student queries. These forums foster a collaborative learning environment where students engage with both the teaching team and their peers. However, the sheer volume of content generated in these forums poses two significant interconnected challenges: How can we effectively identify common misunderstandings that arise in stud…

Juneteenth – what you and I need to know #inspiration #juneteenth https://muz4now.com/2023/juneteenth-wh

Juneteenth - what you and I need to know
Stan Stewart - @muz4now

Efficient Detection of Borromean Linking in Ellipses
Jonathan Strange, Ryan Blair, Alexander R. Klotz
https://arxiv.org/abs/2509.15104 https://arxiv.org/pd…

Efficient Detection of Borromean Linking in Ellipses
We describe a method by which the number of intersections one ellipse makes inside the plane of another can be determined. The method is based on applying a transformation that reverts one ellipse to the unit circle, and examining the intersection points of the transformed second ellipse with the unit circle. This may be used to efficiently determine Hopf linking between two ellipses, or Borromean linking between three ellipses.

Superparamagnetic and Stochastic-Write Magnetic Tunnel Junctions for High-Speed True Random Number Generation in Advanced Computing
Jonathan Z. Sun, Christopher Safranski, Siyuranga Koswata, Pouya Hashemi, Andrew D. Kent
https://arxiv.org/abs/2509.13469

Superparamagnetic and Stochastic-Write Magnetic Tunnel Junctions for High-Speed True Random Number Generation in Advanced Computing
We review two magnetic tunnel junction (MTJ) approaches for compact, low-power, CMOS-integrated true random number generation (TRNG). The first employs passive-read, easy-plane superparamagnetic MTJs (sMTJs) that generate thermal-fluctuation-driven bit streams at $0.5$--$1$~Gb/s per device. The second uses MTJs with magnetically stable free layers, operated with stochastic write pulses to achieve switching probabilities of about $0.5$ (\emph{i.e.}, write error rates of $\simeq 0.5$), achieving …

Adaptive and robust smartphone-based step detection in multiple sclerosis
Lorenza Angelini, Dimitar Stanev, Marta P{\l}onka, Rafa{\l} Klimas, Natan Napi\'orkowski, Gabriela Gonz\'alez Chan, Lisa Bunn, Paul S Glazier, Richard Hosking, Jenny Freeman, Jeremy Hobart, Jonathan Marsden, Licinio Craveiro, Mike D Rinderknecht, Mattia Zanon
https://

Adaptive and robust smartphone-based step detection in multiple sclerosis
Background: Many attempts to validate gait pipelines that process sensor data to detect gait events have focused on the detection of initial contacts only in supervised settings using a single sensor. Objective: To evaluate the performance of a gait pipeline in detecting initial/final contacts using a step detection algorithm adaptive to different test settings, smartphone wear locations, and gait impairment levels. Methods: In GaitLab (ISRCTN15993728), healthy controls (HC) and people with mul…