Search

Activities in underwater environments are paramount in several scenarios, which drives the continuous development of underwater image enhancement techniques. A major challenge in this domain is the depth at which images are captured, with increasing depth resulting in a darker environment. Most existing methods for underwater image enhancement focus on noise removal and color adjustment, with few works dedicated to brightness enhancement. This work introduces a novel unsupervised learning approach to underwater image enhancement using a diffusion model. Our method, called UDBE, is based on conditional diffusion to maintain the brightness details of the unpaired input images. The input image is combined with a color map and a Signal-Noise Relation map (SNR) to ensure stable training and prevent color distortion in the output images. The results demonstrate that our approach achieves an impressive accuracy rate in the datasets UIEB, SUIM and RUIE, well-established underwater image benchmarks. Additionally, the experiments validate the robustness of our approach, regarding the image quality metrics PSNR, SSIM, UIQM, and UISM, indicating the good performance of the brightness enhancement process. The source code is available here: https://github.com/gusanagy/UDBE., Comment: Paper presented at ICMLA 2024

Non-destructive X-ray imaging of thruster parts and assemblies down to the scale of several micrometers is a key technology for electric propulsion research and engineering. It allows for thorough product assurance, rapid state acquisition and implementation of more detailed simulation models to understand the physics of device wear and erosion. Being able to inspect parts as 3D density maps allows insight into inner structures hidden from observation. Generating these density maps and also constructing three dimensional mesh objects for further processing depends on the achievable quality of the reconstruction, which is the inverse of Radon's transformation connecting a stack of projections taken from different angles to the original object's structure. Reconstruction is currently flawed by strong mathematical artifacts induced by the many aligned parts and stark density contrasts commonly found in electric propulsion thrusters., Comment: Presented at the 38th International Electric Propulsion Conference, Toulouse, France

The ion-optic grid-system is the essential part of electrostatic ion thrusters governing performance and lifetime. Therefore reliable measurements of the grid and aperture geometry over the lifetime are necessary to understand and predict the behavior of the system. Many different methods of measurement were introduced over the years to tackle the challenges encountered when diagnosing single electrodes or the whole assembly at once. Modern industrial X-ray micro-computer-tomographs (uCT) offer the possibility to obtain a three-dimensional density map of a grid-system or it's components down to microscopic scales of precision. This information allows a spectrum of new diagnostic opportunities, like complete verification of the manufactured parts against CAD models, detecting internal defects or density-changes or the inspection of the assembled ion-optics and its internal alignment, which is normally prohibited by the lack of optical access to all parts at once. Hence uCT imaging is a promising tool to complement established methods and open up new experimental possibilities, however it also has its own weaknesses and pitfalls. The methods developed for grid-erosion and -geometry measurement of a small state-of-the-art radio-frequency-ion-thruster, the obstacles encountered along the route will be discussed and possible solutions demonstrated., Comment: Presented paper at 37th IEPC in Cambridge, MA

We consider the problem of developing interpretable and computationally efficient matrix decomposition methods for matrices whose entries have bounded support. Such matrices are found in large-scale DNA methylation studies and many other settings. Our approach decomposes the data matrix into a Tucker representation wherein the number of columns in the constituent factor matrices is not constrained. We derive a computationally efficient sampling algorithm to solve for the Tucker decomposition. We evaluate the performance of our method using three criteria: predictability, computability, and stability. Empirical results show that our method has similar performance as other state-of-the-art approaches in terms of held-out prediction and computational complexity, but has significantly better performance in terms of stability to changes in hyper-parameters. The improved stability results in higher confidence in the results in applications where the constituent factors are used to generate and test scientific hypotheses such as DNA methylation analysis of cancer samples., Comment: 33 pages, 18 figures

Given the prompt "Rome is in", can we steer a language model to flip its prediction of an incorrect token "France" to a correct token "Italy" by only multiplying a few relevant activation vectors with scalars? We argue that successfully intervening on a model is a prerequisite for interpreting its internal workings. Concretely, we establish a three-term objective: a successful intervention should flip the correct with the wrong token and vice versa (effectiveness), and leave other tokens unaffected (faithfulness), all while being sparse (minimality). Using gradient-based optimization, this objective lets us learn (and later evaluate) a specific kind of efficient and interpretable intervention: activation scaling only modifies the signed magnitude of activation vectors to strengthen, weaken, or reverse the steering directions already encoded in the model. On synthetic tasks, this intervention performs comparably with steering vectors in terms of effectiveness and faithfulness, but is much more minimal allowing us to pinpoint interpretable model components. We evaluate activation scaling from different angles, compare performance on different datasets, and make activation scalars a learnable function of the activation vectors themselves to generalize to varying-length prompts., Comment: Findings of the Association for Computational Linguistics: EMNLP 2024

Nanoscale magnetic resonance imaging (nanoMRI) aims at obtaining structure at the single molecule level. Most of the techniques for effecting a nanoMRI gradient use small permanent magnets. Here, we present a switchable magnetic field gradient on a tip, which is designed to provide a local and controllable magnetic field with a high gradient on the nanometer scale. We incorporate the gradient field with a nanoscale magnetic resonance sensor, a single nitrogen-vacancy (NV) center in diamond, to provide high-resolution magnetic resonance imaging. The device is a metal microwire deposited along a quartz tip, with the current flowing along the tip inducing a magnetic field around its apex. This field can be manipulated throughout a measurement by controlling the current along the wire. We achieved gradients as high as 1 $\mathrm{\mu}\text{T/nm}$ at fields weaker than 200 $\mathrm{\mu}\text{T}$. Such a gradient can facilitate electron spin mapping with 1 nm resolution using single NV sensors, allowing for nanoscale imaging of electrons. The ability to switch the current on and off and to position the device with high precision overcomes limitations such as limited emitter contrast and the flexibility in sample preparation. Moreover, we show that proximity of the metallic tip to the sensor modifies the Rabi power in a spatially dependent manner, providing regions with enhanced ($\times$3.5) and decreased Rabi power. This spatial gradient, induced by the tip, offers the opportunity for selective pulses on nearby spin species where the same microwave power will result in different spin manipulation characteristics.

Racial and other demographic imputation is necessary for many applications, especially in auditing disparities and outreach targeting in political campaigns. The canonical approach is to construct continuous predictions -- e.g., based on name and geography -- and then to $\textit{discretize}$ the predictions by selecting the most likely class (argmax). We study how this practice produces $\textit{discretization bias}$. In particular, we show that argmax labeling, as used by a prominent commercial voter file vendor to impute race/ethnicity, results in a substantial under-count of African-American voters, e.g., by 28.2% points in North Carolina. This bias can have substantial implications in downstream tasks that use such labels. We then introduce a $\textit{joint optimization}$ approach -- and a tractable $\textit{data-driven thresholding}$ heuristic -- that can eliminate this bias, with negligible individual-level accuracy loss. Finally, we theoretically analyze discretization bias, show that calibrated continuous models are insufficient to eliminate it, and that an approach such as ours is necessary. Broadly, we warn researchers and practitioners against discretizing continuous demographic predictions without considering downstream consequences., Comment: A version of this paper was accepted to the 2024 ACM Conference on Fairness, Accountability, and Transparency

To answer a question, language models often need to integrate prior knowledge learned during pretraining and new information presented in context. We hypothesize that models perform this integration in a predictable way across different questions and contexts: models will rely more on prior knowledge for questions about entities (e.g., persons, places, etc.) that they are more familiar with due to higher exposure in the training corpus, and be more easily persuaded by some contexts than others. To formalize this problem, we propose two mutual information-based metrics to measure a model's dependency on a context and on its prior about an entity: first, the persuasion score of a given context represents how much a model depends on the context in its decision, and second, the susceptibility score of a given entity represents how much the model can be swayed away from its original answer distribution about an entity. We empirically test our metrics for their validity and reliability. Finally, we explore and find a relationship between the scores and the model's expected familiarity with an entity, and provide two use cases to illustrate their benefits., Comment: Long paper accepted at ACL 2024

Detection of moving sources over complicated background is important for several reasons. First is measuring the astrophysical motion of the source. Second is that such motion resulting from atmospheric scintillation, color refraction, or astrophysical reasons is a major source of false alarms for image subtraction methods. We extend the Zackay, Ofek, and Gal-Yam image subtraction formalism to deal with moving sources. The new method, named translient (translational transient) detector, applies hypothesis testing between the hypothesis that the source is stationary and that the source is moving. It can be used to detect source motion or to distinguish between stellar variability and motion. For moving source detection, we show the superiority of translient over the proper image subtraction, using the improvement in the receiver-operating characteristic curve. We show that in the small translation limit, Translient is an optimal detector of point source motion in any direction. Furthermore, it is numerically stable, fast to calculate, and presented in a closed form. Efficient transient detection requires both the proper image subtraction statistics and the translient statistics: when the translient statistic is higher, then the subtraction artifact is likely due to motion. We test our algorithm both on simulated data and on real images obtained by the Large Array Survey Telescope (LAST). We demonstrate the ability of translient to distinguish between motion and variability, which has the potential to reduce the number of false alarms in transients detection. We provide the translient implementation in Python and MATLAB., Comment: 13 pages, 11 figures, submitted to AJ

This paper introduces AL$\ell_0$CORE, a new form of probabilistic non-negative tensor decomposition. AL$\ell_0$CORE is a Tucker decomposition where the number of non-zero elements (i.e., the $\ell_0$-norm) of the core tensor is constrained to a preset value $Q$ much smaller than the size of the core. While the user dictates the total budget $Q$, the locations and values of the non-zero elements are latent variables and allocated across the core tensor during inference. AL$\ell_0$CORE -- i.e., $allo$cated $\ell_0$-$co$nstrained $core$-- thus enjoys both the computational tractability of CP decomposition and the qualitatively appealing latent structure of Tucker. In a suite of real-data experiments, we demonstrate that AL$\ell_0$CORE typically requires only tiny fractions (e.g.,~1%) of the full core to achieve the same results as full Tucker decomposition at only a correspondingly tiny fraction of the cost.

Much of social science is centered around terms like ``ideology'' or ``power'', which generally elude precise definition, and whose contextual meanings are trapped in surrounding language. This paper explores the use of large language models (LLMs) to flexibly navigate the conceptual clutter inherent to social scientific measurement tasks. We rely on LLMs' remarkable linguistic fluency to elicit ideological scales of both legislators and text, which accord closely to established methods and our own judgement. A key aspect of our approach is that we elicit such scores directly, instructing the LLM to furnish numeric scores itself. This approach affords a great deal of flexibility, which we showcase through a variety of different case studies. Our results suggest that LLMs can be used to characterize highly subtle and diffuse manifestations of political ideology in text., Comment: Under review a Harvard Data Science Review. Previously presented at the 4th International Conference of Social Computing in Beijing, China, September 2023, the New Directions in Analyzing Text as Data (TADA) meeting in Amherst, MA, USA, November 2023, and the NeurIPS workshop titled "I Can't Believe It's Not Better!'' Failure Modes in the Age of Foundation Models in New Orleans, LA, December 2023

Many dynamical systems in the real world are naturally described by latent states with intrinsic orderings, such as "ally", "neutral", and "enemy" relationships in international relations. These latent states manifest through countries' cooperative versus conflictual interactions over time. State-space models (SSMs) explicitly relate the dynamics of observed measurements to transitions in latent states. For discrete data, SSMs commonly do so through a state-to-action emission matrix and a state-to-state transition matrix. This paper introduces the Ordered Matrix Dirichlet (OMD) as a prior distribution over ordered stochastic matrices wherein the discrete distribution in the kth row stochastically dominates the (k+1)th, such that probability mass is shifted to the right when moving down rows. We illustrate the OMD prior within two SSMs: a hidden Markov model, and a novel dynamic Poisson Tucker decomposition model tailored to international relations data. We find that models built on the OMD recover interpretable ordered latent structure without forfeiting predictive performance. We suggest future applications to other domains where models with stochastic matrices are popular (e.g., topic modeling), and publish user-friendly code., Comment: Presented at the 26th International Conference on Artificial Intelligence and Statistics (AISTATS) 2023

Measuring the intensity of events is crucial for monitoring and tracking armed conflict. Advances in automated event extraction have yielded massive data sets of "who did what to whom" micro-records that enable data-driven approaches to monitoring conflict. The Goldstein scale is a widely-used expert-based measure that scores events on a conflictual-cooperative scale. It is based only on the action category ("what") and disregards the subject ("who") and object ("to whom") of an event, as well as contextual information, like associated casualty count, that should contribute to the perception of an event's "intensity". This paper takes a latent variable-based approach to measuring conflict intensity. We introduce a probabilistic generative model that assumes each observed event is associated with a latent intensity class. A novel aspect of this model is that it imposes an ordering on the classes, such that higher-valued classes denote higher levels of intensity. The ordinal nature of the latent variable is induced from naturally ordered aspects of the data (e.g., casualty counts) where higher values naturally indicate higher intensity. We evaluate the proposed model both intrinsically and extrinsically, showing that it obtains comparatively good held-out predictive performance., Comment: Long Paper at ACL 2023

We compute the local pro-isomorphic zeta functions at all but finitely many primes for a certain family of class-two-nilpotent Lie lattices of even rank, parametrized by irreducible non-linear polynomials $f(x) \in \mathbb{Z} [x]$, that corresponds to a family of groups introduced by Grunewald and Segal. The result is expressed in terms of a combinatorially defined family of rational functions satisfying a functional equation upon inversion of the variables., Comment: 13 pages, to appear in Proceedings of the American Mathematical Society

Let $L$ be a nilpotent algebra of class two over a compact discrete valuation ring $A$ of characteristic zero or of sufficiently large positive characteristic. Let $q$ be the residue cardinality of $A$. The ideal zeta function of $L$ is a Dirichlet series enumerating finite-index ideals of $L$. We prove that there is a rational function in $q$, $q^m$, $q^{-s}$, and $q^{-ms}$ giving the ideal zeta function of the amalgamation of $m$ copies of $L$ over the derived subring, for every $m \geq 1$, up to an explicit factor. More generally, we prove this for the zeta functions of nilpotent quiver representations of class two defined by Lee and Voll, and in particular for Dirichlet series counting graded submodules of a graded $A$-module. If the algebra $L$, or the quiver representation, is defined over $\mathbb{Z}$, then we obtain a uniform rationality result., Comment: Revised version, to appear in the Bulletin of the London Mathematical Society; 18 pages

We construct infinite families of irreducible supersingular mod $p$ representations of $\mathrm{GL}_2(F)$ with $\mathrm{GL}_2(\mathcal{O}_F)$-socle compatible with Serre's modularity conjecture, where $F / \mathbb{Q}_p$ is any finite extension with residue field $\mathbb{F}_{p^2}$ and ramification degree $e \leq (p-1)/2$. These are the first such examples for ramified $F / \mathbb{Q}_p$., Comment: Some typos corrected, to appear in Israel J. Math