Search

Three-dimensional (3D) high-resolution imaging is essential in microscopy, yet light scattering poses significant challenges in achieving it. Here, we present an approach to holographic imaging of spatially incoherent objects through scattering media, utilizing a virtual medium that replicates the scattering effects of the actual medium. This medium is constructed by retrieving mutually incoherent fields from the object, and exploiting the spatial correlations between them. By numerically propagating the incoherent fields through the virtual medium, we non-invasively compensate for scattering, achieving accurate 3D reconstructions of hidden objects. Experimental validation with fluorescent and synthetic incoherent objects confirms the effectiveness of this approach, opening new possibilities for advanced 3D high-resolution microscopy in scattering environments.

Dense action detection involves detecting multiple co-occurring actions in an untrimmed video while action classes are often ambiguous and represent overlapping concepts. To address this challenge task, we introduce a novel perspective inspired by how humans tackle complex tasks by breaking them into manageable sub-tasks. Instead of relying on a single network to address the entire problem, as in current approaches, we propose decomposing the problem into detecting key concepts present in action classes, specifically, detecting dense static concepts and detecting dense dynamic concepts, and assigning them to distinct, specialized networks. Furthermore, simultaneous actions in a video often exhibit interrelationships, and exploiting these relationships can improve performance. However, we argue that current networks fail to effectively learn these relationships due to their reliance on binary cross-entropy optimization, which treats each class independently. To address this limitation, we propose providing explicit supervision on co-occurring concepts during network optimization through a novel language-guided contrastive learning loss. Our extensive experiments demonstrate the superiority of our approach over state-of-the-art methods, achieving substantial relative improvements of 23.4% and 2.5% mAP on the challenging benchmark datasets, Charades and MultiTHUMOS., Comment: Computer Vision

As LLM agents gain a greater capacity to cause harm, AI developers might increasingly rely on control measures such as monitoring to justify that they are safe. We sketch how developers could construct a "control safety case", which is a structured argument that models are incapable of subverting control measures in order to cause unacceptable outcomes. As a case study, we sketch an argument that a hypothetical LLM agent deployed internally at an AI company won't exfiltrate sensitive information. The sketch relies on evidence from a "control evaluation,"' where a red team deliberately designs models to exfiltrate data in a proxy for the deployment environment. The safety case then hinges on several claims: (1) the red team adequately elicits model capabilities to exfiltrate data, (2) control measures remain at least as effective in deployment, and (3) developers conservatively extrapolate model performance to predict the probability of data exfiltration in deployment. This safety case sketch is a step toward more concrete arguments that can be used to show that a dangerously capable LLM agent is safe to deploy.

Data scarcity in pharmaceutical research has led to reliance on labour-intensive trial-and-error approaches for development rather than data-driven methods. While Machine Learning offers a solution, existing datasets are often small and noisy, limiting their utility. To address this, we developed a Variationally Encoded Conditional Tabular Generative Adversarial Network (VECT-GAN), a novel generative model specifically designed for augmenting small, noisy datasets. We introduce a pipeline where data is augmented before regression model development and demonstrate that this consistently and significantly improves performance over other state-of-the-art tabular generative models. We apply this pipeline across six pharmaceutical datasets, and highlight its real-world applicability by developing novel polymers with medically desirable mucoadhesive properties, which we made and experimentally characterised. Additionally, we pre-train the model on the ChEMBL database of drug-like molecules, leveraging knowledge distillation to enhance its generalisability, making it readily available for use on pharmaceutical datasets containing small molecules, an extremely common pharmaceutical task. We demonstrate the power of synthetic data for regularising small tabular datasets, highlighting its potential to become standard practice in pharmaceutical model development, and make our method, including VECT-GAN pre-trained on ChEMBL available as a pip package., Comment: 30 pages, 6 primary figures, 3 supplementary figures

A remarkable characteristic of quantum computing is the potential for reliable computation despite faulty qubits. This can be achieved through quantum error correction, which is typically implemented by repeatedly applying static syndrome checks, permitting correction of logical information. Recently, the development of time-dynamic approaches to error correction has uncovered new codes and new code implementations. In this work, we experimentally demonstrate three time-dynamic implementations of the surface code, each offering a unique solution to hardware design challenges and introducing flexibility in surface code realization. First, we embed the surface code on a hexagonal lattice, reducing the necessary couplings per qubit from four to three. Second, we walk a surface code, swapping the role of data and measure qubits each round, achieving error correction with built-in removal of accumulated non-computational errors. Finally, we realize the surface code using iSWAP gates instead of the traditional CNOT, extending the set of viable gates for error correction without additional overhead. We measure the error suppression factor when scaling from distance-3 to distance-5 codes of $\Lambda_{35,\text{hex}} = 2.15(2)$, $\Lambda_{35,\text{walk}} = 1.69(6)$, and $\Lambda_{35,\text{iSWAP}} = 1.56(2)$, achieving state-of-the-art error suppression for each. With detailed error budgeting, we explore their performance trade-offs and implications for hardware design. This work demonstrates that dynamic circuit approaches satisfy the demands for fault-tolerance and opens new alternative avenues for scalable hardware design., Comment: 11 pages, 5 figures, Supplementary Information

Quantum error correction is essential for bridging the gap between the error rates of physical devices and the extremely low logical error rates required for quantum algorithms. Recent error-correction demonstrations on superconducting processors have focused primarily on the surface code, which offers a high error threshold but poses limitations for logical operations. In contrast, the color code enables much more efficient logic, although it requires more complex stabilizer measurements and decoding techniques. Measuring these stabilizers in planar architectures such as superconducting qubits is challenging, and so far, realizations of color codes have not addressed performance scaling with code size on any platform. Here, we present a comprehensive demonstration of the color code on a superconducting processor, achieving logical error suppression and performing logical operations. Scaling the code distance from three to five suppresses logical errors by a factor of $\Lambda_{3/5}$ = 1.56(4). Simulations indicate this performance is below the threshold of the color code, and furthermore that the color code may be more efficient than the surface code with modest device improvements. Using logical randomized benchmarking, we find that transversal Clifford gates add an error of only 0.0027(3), which is substantially less than the error of an idling error correction cycle. We inject magic states, a key resource for universal computation, achieving fidelities exceeding 99% with post-selection (retaining about 75% of the data). Finally, we successfully teleport logical states between distance-three color codes using lattice surgery, with teleported state fidelities between 86.5(1)% and 90.7(1)%. This work establishes the color code as a compelling research direction to realize fault-tolerant quantum computation on superconducting processors in the near future.

Recent latent-space monitoring techniques have shown promise as defenses against LLM attacks. These defenses act as scanners that seek to detect harmful activations before they lead to undesirable actions. This prompts the question: Can models execute harmful behavior via inconspicuous latent states? Here, we study such obfuscated activations. We show that state-of-the-art latent-space defenses -- including sparse autoencoders, representation probing, and latent OOD detection -- are all vulnerable to obfuscated activations. For example, against probes trained to classify harmfulness, our attacks can often reduce recall from 100% to 0% while retaining a 90% jailbreaking rate. However, obfuscation has limits: we find that on a complex task (writing SQL code), obfuscation reduces model performance. Together, our results demonstrate that neural activations are highly malleable: we can reshape activation patterns in a variety of ways, often while preserving a network's behavior. This poses a fundamental challenge to latent-space defenses., Comment: Project page: https://obfuscated-activations.github.io/ Code: https://github.com/LukeBailey181/obfuscated-activations

The "miscentering effect," i.e., the offset between a galaxy cluster's optically-defined center and the center of its gravitational potential, is a significant systematic effect on brightest cluster galaxy (BCG) studies and cluster lensing analyses. We perform a cross-match between the optical cluster catalog from the Hyper Suprime-Cam (HSC) Survey S19A Data Release and the Sunyaev-Zeldovich cluster catalog from Data Release 5 of the Atacama Cosmology Telescope (ACT). We obtain a sample of 186 clusters in common in the redshift range $0.1 \leq z \leq 1.4$ over an area of 469 deg$^2$. By modeling the distribution of centering offsets in this fiducial sample, we find a miscentered fraction (corresponding to clusters offset by more than 330 kpc) of ~25%, a value consistent with previous miscentering studies. We examine the image of each miscentered cluster in our sample and identify one of several reasons to explain the miscentering. Some clusters show significant miscentering for astrophysical reasons, i.e., ongoing cluster mergers. Others are miscentered due to non-astrophysical, systematic effects in the HSC data or the cluster-finding algorithm. After removing all clusters with clear, non-astrophysical causes of miscentering from the sample, we find a considerably smaller miscentered fraction, ~10%. We show that the gravitational lensing signal within 1 Mpc of miscentered clusters is considerably smaller than that of well-centered clusters, and we suggest that the ACT SZ centers are a better estimate of the true cluster potential centroid., Comment: 20 pages, 15 figures. Accepted for publication in MNRAS

The precise characterization and mitigation of systematic effects is one of the biggest roadblocks impeding the detection of the fluctuations of cosmological 21cm signals. Missing data in radio cosmological experiments, often due to radio frequency interference (RFI), poses a particular challenge to power spectrum analysis as it could lead to the ringing of bright foreground modes in Fourier space, heavily contaminating the cosmological signals. Here we show that the problem of missing data becomes even more arduous in the presence of systematic effects. Using a realistic numerical simulation, we demonstrate that partially flagged data combined with systematic effects can introduce significant foreground ringing. We show that such an effect can be mitigated through inpainting the missing data. We present a rigorous statistical framework that incorporates the process of inpainting missing data into a quadratic estimator of the 21cm power spectrum. Under this framework, the uncertainties associated with our inpainting method and its impact on power spectrum statistics can be understood. These results are applied to the latest Phase II observations taken by the Hydrogen Epoch of Reionization Array, forming a crucial component in power spectrum analyses as we move toward detecting 21cm signals in the ever more noisy RFI environment., Comment: 25 pages, 11 figures; Replaced to match accepted ApJ version. New version contains small editorial changes throughout in response to referee comments, no changes to results

Frontier artificial intelligence (AI) systems pose increasing risks to society, making it essential for developers to provide assurances about their safety. One approach to offering such assurances is through a safety case: a structured, evidence-based argument aimed at demonstrating why the risk associated with a safety-critical system is acceptable. In this article, we propose a safety case template for offensive cyber capabilities. We illustrate how developers could argue that a model does not have capabilities posing unacceptable cyber risks by breaking down the main claim into progressively specific sub-claims, each supported by evidence. In our template, we identify a number of risk models, derive proxy tasks from the risk models, define evaluation settings for the proxy tasks, and connect those with evaluation results. Elements of current frontier safety techniques - such as risk models, proxy tasks, and capability evaluations - use implicit arguments for overall system safety. This safety case template integrates these elements using the Claims Arguments Evidence (CAE) framework in order to make safety arguments coherent and explicit. While uncertainties around the specifics remain, this template serves as a proof of concept, aiming to foster discussion on AI safety cases and advance AI assurance., Comment: 21 pages, 6 figures, 1 table

The high electron temperature in galaxy clusters ($>1\,$keV or $>10^7\,$K) leads to corrections at the level of a few percent in their thermal Sunyaev-Zeldovich effect signatures. Both the size and frequency dependence of these corrections, which are known as relativistic temperature corrections, depend upon the temperature of the objects. In this work we exploit this effect to measure the average temperature of a stack of Compton-$y$ selected clusters. Specifically, we apply the "spectroscopic method" and search for the temperature that best fits the clusters' signal measured at frequencies from 30 to 545 GHz by the Atacama Cosmology Telescope and Planck satellite. We measure the average temperature of clusters detected in the ACT maps to be $8.5\pm 2.4\,$keV, with an additional systematic error of comparable amplitude dominated by passband uncertainty. Upcoming surveys, such as the Simons Observatory and CMB-S4, have the potential to dramatically improve upon these measurements and thereby enable precision studies of cluster temperatures with millimeter observations. The key challenge for future observations will be mitigating instrumental systematic effects, which already limit this analysis., Comment: 21 pages with 17 figures

According to CMB measurements, baryonic matter constitutes about $5\%$ of the mass-energy density of the universe. A significant population of these baryons, for a long time referred to as `missing', resides in a low density, warm-hot intergalactic medium (WHIM) outside galaxy clusters, tracing the ``cosmic web'', a network of large scale dark matter filaments. Various studies have detected this inter-cluster gas, both by stacking and by observing individual filaments in compact, massive systems. In this paper, we study short filaments (< 10 Mpc) connecting massive clusters ($M_{500} \approx 3\times 10^{14} M_{\odot}$) detected by the Atacama Cosmology Telescope (ACT) using the scattering of CMB light off the ionised gas, a phenomenon known as the thermal Sunyaev-Zeldovich (tSZ) effect. The first part of this work is a search for suitable candidates for high resolution follow-up tSZ observations. We identify four cluster pairs with an intercluster signal above the noise floor (S/N $>$ 2), including two with a tentative $>2\sigma$ statistical significance for an intercluster bridge from the ACT data alone. In the second part of this work, starting from the same cluster sample, we directly stack on ${\sim}100$ cluster pairs and observe an excess SZ signal between the stacked clusters of $y=(7.2^{+2.3}_{-2.5})\times 10^{-7}$ with a significance of $3.3\sigma$. It is the first tSZ measurement of hot gas between clusters in this range of masses at moderate redshift ($\langle z\rangle\approx 0.5$). We compare this to the signal from simulated cluster pairs with similar redshifts and separations in the THE300 and MAGNETICUM Pathfinder cosmological simulations and find broad consistency. Additionally, we show that our measurement is consistent with scaling relations between filament parameters and mass of the embedded halos identified in simulations., Comment: 37 pages, 17 images

Recent illumination estimation methods have focused on enhancing the resolution and improving the quality and diversity of the generated textures. However, few have explored tailoring the neural network architecture to the Equirectangular Panorama (ERP) format utilised in image-based lighting. Consequently, high dynamic range images (HDRI) results usually exhibit a seam at the side borders and textures or objects that are warped at the poles. To address this shortcoming we propose a novel architecture, 360U-Former, based on a U-Net style Vision-Transformer which leverages the work of PanoSWIN, an adapted shifted window attention tailored to the ERP format. To the best of our knowledge, this is the first purely Vision-Transformer model used in the field of illumination estimation. We train 360U-Former as a GAN to generate HDRI from a limited field of view low dynamic range image (LDRI). We evaluate our method using current illumination estimation evaluation protocols and datasets, demonstrating that our approach outperforms existing and state-of-the-art methods without the artefacts typically associated with the use of the ERP format., Comment: Accepted at AIM Workshop 2024 at ECCV 2024, 18 pages, 6 figures

We consider the problem of low probability estimation: given a machine learning model and a formally-specified input distribution, how can we estimate the probability of a binary property of the model's output, even when that probability is too small to estimate by random sampling? This problem is motivated by the need to improve worst-case performance, which distribution shift can make much more likely. We study low probability estimation in the context of argmax sampling from small transformer language models. We compare two types of methods: importance sampling, which involves searching for inputs giving rise to the rare output, and activation extrapolation, which involves extrapolating a probability distribution fit to the model's logits. We find that importance sampling outperforms activation extrapolation, but both outperform naive sampling. Finally, we explain how minimizing the probability estimate of an undesirable behavior generalizes adversarial training, and argue that new methods for low probability estimation are needed to provide stronger guarantees about worst-case performance., Comment: 29 pages, 9 figures

One of the most challenging problems in the computational study of localization in quantum manybody systems is to capture the effects of rare events, which requires sampling over exponentially many disorder realizations. We implement an efficient procedure on a quantum processor, leveraging quantum parallelism, to efficiently sample over all disorder realizations. We observe localization without disorder in quantum many-body dynamics in one and two dimensions: perturbations do not diffuse even though both the generator of evolution and the initial states are fully translationally invariant. The disorder strength as well as its density can be readily tuned using the initial state. Furthermore, we demonstrate the versatility of our platform by measuring Renyi entropies. Our method could also be extended to higher moments of the physical observables and disorder learning.

Accurate state estimation is crucial for legged robot locomotion, as it provides the necessary information to allow control and navigation. However, it is also challenging, especially in scenarios with uneven and slippery terrain. This paper presents a new Invariant Extended Kalman filter for legged robot state estimation using only proprioceptive sensors. We formulate the methodology by combining recent advances in state estimation theory with the use of robust cost functions in the measurement update. We tested our methodology on quadruped robots through experiments and public datasets, showing that we can obtain a pose drift up to 40% lower in trajectories covering a distance of over 450m, in comparison with a state-of-the-art Invariant Extended Kalman filter., Comment: Accepted to the IEEE-RAS International Conference on Humanoid Robots 2024

Christiano et al. (2022) define a *heuristic estimator* to be a hypothetical algorithm that estimates the values of mathematical expressions from arguments. In brief, a heuristic estimator $\mathbb{G}$ takes as input a mathematical expression $Y$ and a formal "heuristic argument" $\pi$, and outputs an estimate $\mathbb{G}(Y \mid \pi)$ of $Y$. In this work, we argue for the informal principle that a heuristic estimator ought not to be able to predict its own errors, and we explore approaches to formalizing this principle. Most simply, the principle suggests that $\mathbb{G}(Y - \mathbb{G}(Y \mid \pi) \mid \pi)$ ought to equal zero for all $Y$ and $\pi$. We argue that an ideal heuristic estimator ought to satisfy two stronger properties in this vein, which we term *iterated estimation* (by analogy to the law of iterated expectations) and *error orthogonality*. Although iterated estimation and error orthogonality are intuitively appealing, it can be difficult to determine whether a given heuristic estimator satisfies the properties. As an alternative approach, we explore *accuracy*: a property that (roughly) states that $\mathbb{G}$ has zero average error over a distribution of mathematical expressions. However, in the context of two estimation problems, we demonstrate barriers to creating an accurate heuristic estimator. We finish by discussing challenges and potential paths forward for finding a heuristic estimator that accords with our intuitive understanding of how such an estimator ought to behave, as well as the potential applications of heuristic estimators to understanding the behavior of neural networks., Comment: 47 pages, 2 tables, 1 figure

The Robert Stobie Spectrograph (RSS) on the Southern African Large Telescope (SALT) offers multi-object spectroscopy over an 8' field-of-view at resolutions up to 3000. Reduction is typically conducted using RSSMOSPipeline, which performs basic data calibrations, sky subtraction, and wavelength calibration. However, flux calibration of SALT-RSS using spectrophotometric standard star observations is difficult due to variable primary mirror illumination. We describe a novel approach where stars with Sloan Digital Sky Survey spectra are included as alignment stars on RSS slitmasks and then used to perform flux calibration of the resulting data. RSS offers multiple settings that can be pieced together to cover the entire optical range, utilizing grating angle dithers to fill chip gaps. We introduce a non-linear reprojection routine that defines an exponential wavelength array spanning 3500-9500 Angstroms with gradually decreasing resolution and then reprojects several individual settings into a single 2D spectrum for each object. Our flux calibration and non-linear reprojection routines are released as part of the Calibration And Reprojection for RSS Pipeline (CARRSSPipeline; https://github.com/GeorgeTheGeorgian/CARRSSPipeline.git ), that enables the extraction of full-optical-coverage, flux-calibrated, medium-resolution one-dimensional spectra., Comment: 12 pages, 9 figures, and 4 tables. Submitted to PASP

Lattice gauge theories (LGTs) can be employed to understand a wide range of phenomena, from elementary particle scattering in high-energy physics to effective descriptions of many-body interactions in materials. Studying dynamical properties of emergent phases can be challenging as it requires solving many-body problems that are generally beyond perturbative limits. We investigate the dynamics of local excitations in a $\mathbb{Z}_2$ LGT using a two-dimensional lattice of superconducting qubits. We first construct a simple variational circuit which prepares low-energy states that have a large overlap with the ground state; then we create particles with local gates and simulate their quantum dynamics via a discretized time evolution. As the effective magnetic field is increased, our measurements show signatures of transitioning from deconfined to confined dynamics. For confined excitations, the magnetic field induces a tension in the string connecting them. Our method allows us to experimentally image string dynamics in a (2+1)D LGT from which we uncover two distinct regimes inside the confining phase: for weak confinement the string fluctuates strongly in the transverse direction, while for strong confinement transverse fluctuations are effectively frozen. In addition, we demonstrate a resonance condition at which dynamical string breaking is facilitated. Our LGT implementation on a quantum processor presents a novel set of techniques for investigating emergent particle and string dynamics.

Precision timekeeping is fundamental to modern technologies such as Global Navigation Satellite Systems (GNSS), communication networks, financial transactions, and power grid management. Over the past 50 years, microwave atomic clocks have been the standard for timing precision. The new generation of optical atomic clocks have demonstrated orders of magnitude better performance, and are now transitioning from research to practical applications. We provide a web resource that tracks the performance of these optical atomic clocks, measured in terms of their Allan deviation at various integration times, against their SWaP requirements via an interactive plot. The most current data and additional resources are available online, providing a continuously updated reference for those interested in precision timing., Comment: 5 pages, 2 figures. For more information see https://a1120960.github.io/PAC-SWaP/

We conduct a systematic search for astrophysical transients using data from the Atacama Cosmology Telescope (ACT). The data were taken from 2017 to 2022 in three frequency bands spanning 77 GHz to 277 GHz. In this paper we present a pipeline for transient detection using single observation maps where each pixel of a map contains one observation with an integration time of approximately four minutes. We find 34 transient events at 27 unique locations. All but two of the transients are associated with Galactic stars and exhibit a wide range of properties. We also detect an event coincident with the classical nova, YZ Ret and one event consistent with a flaring active galactic nucleus. We notably do not detect any reverse shock emission from gamma ray bursts, a non-detection which is in tension with current models., Comment: 23 pages, 10 figures, 10 tables. First and second author share equal contributions. Article and accompanying data submitted to ApJ. Data tables will be made available upon publication

We introduce a formal notion of defendability against backdoors using a game between an attacker and a defender. In this game, the attacker modifies a function to behave differently on a particular input known as the "trigger", while behaving the same almost everywhere else. The defender then attempts to detect the trigger at evaluation time. If the defender succeeds with high enough probability, then the function class is said to be defendable. The key constraint on the attacker that makes defense possible is that the attacker's strategy must work for a randomly-chosen trigger. Our definition is simple and does not explicitly mention learning, yet we demonstrate that it is closely connected to learnability. In the computationally unbounded setting, we use a voting algorithm of Hanneke et al. (2022) to show that defendability is essentially determined by the VC dimension of the function class, in much the same way as PAC learnability. In the computationally bounded setting, we use a similar argument to show that efficient PAC learnability implies efficient defendability, but not conversely. On the other hand, we use indistinguishability obfuscation to show that the class of polynomial size circuits is not efficiently defendable. Finally, we present polynomial size decision trees as a natural example for which defense is strictly easier than learning. Thus, we identify efficient defendability as a notable intermediate concept in between efficient learnability and obfuscation., Comment: 29 pages

Using a sample of 2800 galaxy clusters identified in the Dark Energy Survey across the redshift range $0.20 < z < 0.60$, we characterize the hierarchical assembly of Bright Central Galaxies (BCGs) and the surrounding intracluster light (ICL). To quantify hierarchical formation we use the stellar mass - halo mass (SMHM) relation for the BCG+ICL system and incorporate the magnitude gap (M14), the difference in brightness between the BCG (measured within 30kpc) and 4th brightest cluster member galaxy within 0.5 $R_{200,c}$. The inclusion of M14, which traces BCG hierarchical growth, increases the slope and decreases the intrinsic scatter in the SMHM relation, highlighting that it is a latent variable within the BCG+ICL SMHM relation. Moreover, the correlation with M14 decreases at large radii from the BCG's centre. However, the stellar light within the BCG+ICL transition region (30kpc - 80kpc) most strongly correlates with the dark matter halo mass and has a statistically significant correlation with M14. As the light in the transition region and M14 are independent measurements, the transition region may grow as a result of the BCG's hierarchical two-phase formation. Additionally, as M14 and ICL result from hierarchical growth, we use a stacked sample and find that clusters with large M14 values are characterized by larger ICL and BCG+ICL fractions, which illustrates that the merger processes that build the BCG stellar mass also grow the ICL. Furthermore, this may suggest that M14 combined with the ICL fraction can be used as a method to identify dynamically relaxed clusters., Comment: 16 pages, 5 Figures, submitted to MNRAS on 8/30/2024