Search

A common paradigm for improving fuzzing performance is to focus on selected regions of a program rather than its entirety. While previous work has largely explored how these locations can be reached, their selection, that is, the where, has received little attention so far. A common paradigm for improving fuzzing performance is to focus on selected regions of a program rather than its entirety. While previous work has largely explored how these locations can be reached, their selection, that is, the where, has received little attention so far. In this paper, we fill this gap and present the first comprehensive analysis of target selection methods for fuzzing. To this end, we examine papers from leading security and software engineering conferences, identifying prevalent methods for choosing targets. By modeling these methods as general scoring functions, we are able to compare and measure their efficacy on a corpus of more than 1,600 crashes from the OSS-Fuzz project. Our analysis provides new insights for target selection in practice: First, we find that simple software metrics significantly outperform other methods, including common heuristics used in directed fuzzing, such as recently modified code or locations with sanitizer instrumentation. Next to this, we identify language models as a promising choice for target selection. In summary, our work offers a new perspective on directed fuzzing, emphasizing the role of target selection as an orthogonal dimension to improve performance.

Datasets play a critical role in medical imaging research, yet issues such as label quality, shortcuts, and metadata are often overlooked. This lack of attention may harm the generalizability of algorithms and, consequently, negatively impact patient outcomes. While existing medical imaging literature reviews mostly focus on machine learning (ML) methods, with only a few focusing on datasets for specific applications, these reviews remain static -- they are published once and not updated thereafter. This fails to account for emerging evidence, such as biases, shortcuts, and additional annotations that other researchers may contribute after the dataset is published. We refer to these newly discovered findings of datasets as research artifacts. To address this gap, we propose a living review that continuously tracks public datasets and their associated research artifacts across multiple medical imaging applications. Our approach includes a framework for the living review to monitor data documentation artifacts, and an SQL database to visualize the citation relationships between research artifact and dataset. Lastly, we discuss key considerations for creating medical imaging datasets, review best practices for data annotation, discuss the significance of shortcuts and demographic diversity, and emphasize the importance of managing datasets throughout their entire lifecycle. Our demo is publicly available at http://130.226.140.142., Comment: Manuscript under review

Scalar particles traveling faster than a subluminal gravitational wave generate gravitons via gravitational Cherenkov radiation. In this paper, we investigate graviton production by the primordial plasma within the framework of modified gravity in the early Universe, generating a relic graviton background. We find that for the minimal model, where only the speed of gravitational waves is modified and a standard model plasma minimally couples to gravity, the relic graviton background can be enhanced by several orders of magnitude, but still agrees with the Big Bang Nucleosynthesis (BBN) bound in most cases. Moreover, we also find that for Horndeski theories, such as Galileon theory, the relic background produced by the thermalized scalar field can reach significant amplitudes, exceeding the BBN bound for a region of the parameter space. By requiring the relic graviton background to remain consistent with the BBN constraint, we derive limits on the gravitational wave speed at early times in these modified gravity theories., Comment: 13 pages, 5 figures

Purpose: Reliable image quality assessment is crucial for evaluating new motion correction methods for magnetic resonance imaging. In this work, we compare the performance of commonly used reference-based and reference-free image quality metrics on a unique dataset with real motion artifacts. We further analyze the image quality metrics' robustness to typical pre-processing techniques. Methods: We compared five reference-based and five reference-free image quality metrics on data acquired with and without intentional motion (2D and 3D sequences). The metrics were recalculated seven times with varying pre-processing steps. The anonymized images were rated by radiologists and radiographers on a 1-5 Likert scale. Spearman correlation coefficients were computed to assess the relationship between image quality metrics and observer scores. Results: All reference-based image quality metrics showed strong correlation with observer assessments, with minor performance variations across sequences. Among reference-free metrics, Average Edge Strength offers the most promising results, as it consistently displayed stronger correlations across all sequences compared to the other reference-free metrics. Overall, the strongest correlation was achieved with percentile normalization and restricting the metric values to the skull-stripped brain region. In contrast, correlations were weaker when not applying any brain mask and using min-max or no normalization. Conclusion: Reference-based metrics reliably correlate with radiological evaluation across different sequences and datasets. Pre-processing steps, particularly normalization and brain masking, significantly influence the correlation values. Future research should focus on refining pre-processing techniques and exploring machine learning approaches for automated image quality evaluation.

Hyperpolarized nanoparticles (NPs) offer high polarization levels with room temperature relaxation times exceeding half an hour. In this work, we demonstrate that the achievable hyperpolarization enhancement and relaxation (decay) time at room temperature are largely independent of the particle size contrary to previous assumptions. This is explained through first-principles spin-diffusion coefficient calculations and finite-element polarization simulations. The simulated zero-quantum (flip-flop) line width governing the spin diffusion is found to agree with the experimentally accessible single-quantum (single spin flip, e.g. radio-frequency pulse) line width. The transport of hyperpolarization from strongly hyperfine-coupled spins towards the bulk is most likelybelieved to be responsible for the slow polarization dynamics including long room temperature decay time. The line width and spin-diffusion simulations are extended to other cubic crystal structures and analytical expressions, which only require insertion of the gyromagnetic ratio, lattice constant, isotope abundance and measured spectral density distribution (nuclear line width), are fitted. The presented simulations can be adjusted to study spin diffusion in other materials.

Vision-Language Models (VLMs) excel in diverse visual tasks but face challenges in document understanding, which requires fine-grained text processing. While typical visual tasks perform well with low-resolution inputs, reading-intensive applications demand high-resolution, resulting in significant computational overhead. Using OCR-extracted text in VLM prompts partially addresses this issue but underperforms compared to full-resolution counterpart, as it lacks the complete visual context needed for optimal performance. We introduce DocVLM, a method that integrates an OCR-based modality into VLMs to enhance document processing while preserving original weights. Our approach employs an OCR encoder to capture textual content and layout, compressing these into a compact set of learned queries incorporated into the VLM. Comprehensive evaluations across leading VLMs show that DocVLM significantly reduces reliance on high-resolution images for document understanding. In limited-token regimes (448$\times$448), DocVLM with 64 learned queries improves DocVQA results from 56.0% to 86.6% when integrated with InternVL2 and from 84.4% to 91.2% with Qwen2-VL. In LLaVA-OneVision, DocVLM achieves improved results while using 80% less image tokens. The reduced token usage allows processing multiple pages effectively, showing impressive zero-shot results on DUDE and state-of-the-art performance on MP-DocVQA, highlighting DocVLM's potential for applications requiring high-performance and efficiency.

Foundation models (FMs), i.e., models trained on a vast amount of typically unlabeled data, have become popular and available recently for the domain of histopathology. The key idea is to extract semantically rich vectors from any input patch, allowing for the use of simple subsequent classification networks potentially reducing the required amounts of labeled data, and increasing domain robustness. In this work, we investigate to which degree this also holds for mitotic figure classification. Utilizing two popular public mitotic figure datasets, we compared linear probing of five publicly available FMs against models trained on ImageNet and a simple ResNet50 end-to-end-trained baseline. We found that the end-to-end-trained baseline outperformed all FM-based classifiers, regardless of the amount of data provided. Additionally, we did not observe the FM-based classifiers to be more robust against domain shifts, rendering both of the above assumptions incorrect., Comment: 6 pages, 2 figures

We propose an extended demand response program, based on ancillary service for supplying flexible electricity demand. In our proposed scheme, we suggest a broader management model to control the scheduling and power consumption of Bitcoin mining machines. The main aspect that we focus on is suppressing the power ramping and related transient effects. We extend previous works on the subject, that study the impact of incorporating cryptocurrency mining machines into existing power grid, and explore the potential profit of exploiting this flexible load in the Israeli electricity market. We analyze a trend based on historical data, of increasing electricity prices and ramping costs due to the increasing penetration of renewable energy sources. We suggest an extension to the unit commitment problem from which we obtain the scheduling scheme of the Bitcoin mining machines. We use simulation and the real-world data acquired from the "Noga" grid operator to verify the proposed ancillary service and test its practical limits for reducing the ramping costs, under changing ratio of energy production from renewable sources. Out results suggests that the machine price and ratio of production from renewable sources plays a significant role in determining the profitability of the proposed demand-response program., Comment: 8 pages, 7 figures

Vision-Language (VL) models have garnered considerable research interest; however, they still face challenges in effectively handling text within images. To address this limitation, researchers have developed two approaches. The first method involves utilizing external Optical Character Recognition (OCR) tools to extract textual information from images, which is then prepended to other textual inputs. The second strategy focuses on employing extremely high-resolution images to improve text recognition capabilities. In this paper, we focus on enhancing the first strategy by introducing a novel method, named TAP-VL, which treats OCR information as a distinct modality and seamlessly integrates it into any VL model. TAP-VL employs a lightweight transformer-based OCR module to receive OCR with layout information, compressing it into a short fixed-length sequence for input into the LLM. Initially, we conduct model-agnostic pretraining of the OCR module on unlabeled documents, followed by its integration into any VL architecture through brief fine-tuning. Extensive experiments demonstrate consistent performance improvements when applying TAP-VL to top-performing VL models, across scene-text and document-based VL benchmarks.

Artificial intelligence (AI)-based decision support systems hold promise for enhancing diagnostic accuracy and efficiency in computational pathology. However, human-AI collaboration can introduce and amplify cognitive biases, such as confirmation bias caused by false confirmation when erroneous human opinions are reinforced by inaccurate AI output. This bias may worsen when time pressure, ubiquitously present in routine pathology, strains practitioners' cognitive resources. We quantified confirmation bias triggered by AI-induced false confirmation and examined the role of time constraints in a web-based experiment, where trained pathology experts (n=28) estimated tumor cell percentages. Our results suggest that AI integration may fuel confirmation bias, evidenced by a statistically significant positive linear-mixed-effects model coefficient linking AI recommendations mirroring flawed human judgment and alignment with system advice. Conversely, time pressure appeared to weaken this relationship. These findings highlight potential risks of AI use in healthcare and aim to support the safe integration of clinical decision support systems.

Artificial intelligence (AI)-based clinical decision support systems (CDSS) promise to enhance diagnostic accuracy and efficiency in computational pathology. However, human-AI collaboration might introduce automation bias, where users uncritically follow automated cues. This bias may worsen when time pressure strains practitioners' cognitive resources. We quantified automation bias by measuring the adoption of negative system consultations and examined the role of time pressure in a web-based experiment, where trained pathology experts (n=28) estimated tumor cell percentages. Our results indicate that while AI integration led to a statistically significant increase in overall performance, it also resulted in a 7% automation bias rate, where initially correct evaluations were overturned by erroneous AI advice. Conversely, time pressure did not exacerbate automation bias occurrence, but appeared to increase its severity, evidenced by heightened reliance on the system's negative consultations and subsequent performance decline. These findings highlight potential risks of AI use in healthcare.

Natural language processing (NLP) in the medical domain can underperform in real-world applications involving small datasets in a non-English language with few labeled samples and imbalanced classes. There is yet no consensus on how to approach this problem. We evaluated a set of NLP models including BERT-like transformers, few-shot learning with sentence transformers (SetFit), and prompted large language models (LLM), using three datasets of radiology reports on magnetic resonance images of epilepsy patients in Danish, a low-resource language. Our results indicate that BERT-like models pretrained in the target domain of radiology reports currently offer the optimal performances for this scenario. Notably, the SetFit and LLM models underperformed compared to BERT-like models, with LLM performing the worst. Importantly, none of the models investigated was sufficiently accurate to allow for text classification without any supervision. However, they show potential for data filtering, which could reduce the amount of manual labeling required.

Nearly two billion children across the globe are growing up in an increasingly digital world. Yet even as the ways that kids engage with new technologies constantly evolve, our efforts to attend to their healthy development hasn't kept pace. It's time to prioritize the well-being of children in digital spaces, which we believe requires a collaborative effort across academia, industry, the policy community, and practitioners to understand how the design of digital products and services can advance children's best interests. One framework that addresses this gap was created by the Responsible Innovation in Technology for Children initiative (RITEC), developed by UNICEF and the LEGO Group and supported by the LEGO Foundation. Initially published in 2022, the framework is based on syntheses of developmental research, qualitative analyses of conversations with children, and quantitative analyses from survey data spanning 30 countries (UNICEF, 2022). It includes eight components: "competence," "creativity," "diversity, equity, and inclusion," "emotional regulation," "empowerment," "safety and security," "self-actualization," and "social connection." Each component is multidimensional and draws from developmental and cognitive science. To bring these ideas to life and to elaborate on their connections to digital design, the Joan Ganz Cooney Center has created this guide for developers of interactive products for children. We provide definitions and summaries from related research and offer recommendations for making digital experiences and products for children of all ages. We hope the guide will spark conversations among creators of media and technology for children, inspire them to consider how their products can support children's well-being, and take action!

Joint Energy Models (JEMs), while drawing significant research attention, have not been successfully scaled to real-world, high-resolution datasets. We present EB-CLIP, a novel approach extending JEMs to the multimodal vision-language domain using CLIP, integrating both generative and discriminative objectives. For the generative objective, we introduce an image-text joint-energy function based on Cosine similarity in the CLIP space, training CLIP to assign low energy to real image-caption pairs and high energy otherwise. For the discriminative objective, we employ contrastive adversarial loss, extending the adversarial training objective to the multimodal domain. EB-CLIP not only generates realistic images from text but also achieves competitive results on the compositionality benchmark, outperforming leading methods with fewer parameters. Additionally, we demonstrate the superior guidance capability of EB-CLIP by enhancing CLIP-based generative frameworks and converting unconditional diffusion models to text-based ones. Lastly, we show that EB-CLIP can serve as a more robust evaluation metric for text-to-image generative tasks than CLIP.

We discuss the non-linear interactions within the VCDM model, a type II minimally modified gravity model with the same number of degrees of freedom as in General Relativity but not connected to the latter by field redefinitions. During an inflationary phase in the early universe, if the VCDM potential does not modify the slow-roll behavior of the inflaton field, we recover, up to the leading order, the standard results for the bispectrum in slow-roll inflation. On the other hand, if the VCDM potential becomes dominant, the interactions can strongly deviate and even violate the Maldacena's consistency relation for the local non-Gaussianities. Furthermore, we apply the formalism to the recently introduced bouncing model in VCDM, and show that the bispectrum still respects current observational constraints. Future measurements on non-Gaussianities of the local type should provide the test ground for the model's validity., Comment: 22 pages, Prepared for a special issue of the International Journal of Modern Physics A, which is devoted to the memory of V.A.Rubakov

We study secondary gravitational wave production in Horndenski gravity, when the scalar field dominates the very early universe. We find that higher derivative interactions easily dominate the source term on subhorizon scales and significantly enhance the amplitude of induced GWs. We analytically derive, for the first time, the Horndeski induced GW spectrum for a general class of power-law solutions. The main effect of modifications of gravity are stronger resonances and a growth of tensor fluctuations on small scales. The maximum attainable amplitude of the induced GW spectrum is bounded by the possible backreaction of higher derivatives on curvature fluctuations, thereby shutting down the source term to induced GWs. We argue that the maximum attainable amplitude depends linearly on the primordial curvature spectrum ($\Omega_{\rm GW}\propto {\cal P}_\zeta$), as opposed to the standard case where it depends quadratically. Resonances may further enhance the maximum amplitude by a factor $(k/{\cal H}_t)^2$ or $(k/{\cal H}_t)$ respectively for sharp and broad peaks (including a scale invariant) primordial spectrum, where ${\cal H}_t$ is the comoving horizon at the time when standard gravity is recovered. Remarkably, in the scale invariant case, the Horndeski-induced GW spectrum grows as $k^3$. This opens up the interesting possibility that induced GWs might be observable despite no enhancement of the primordial curvature spectrum. Our formalism can be generalized to a wider class of solutions and to more general scalar-tensor theories, such as DHOST and spatially covariant gravity. In the appendices we provide discussions on the gauge issue and disformal transformations of induced GWs., Comment: 66 pages, 8 figures

The count of mitotic figures (MFs) observed in hematoxylin and eosin (H&E)-stained slides is an important prognostic marker as it is a measure for tumor cell proliferation. However, the identification of MFs has a known low inter-rater agreement. Deep learning algorithms can standardize this task, but they require large amounts of annotated data for training and validation. Furthermore, label noise introduced during the annotation process may impede the algorithm's performance. Unlike H&E, the mitosis-specific antibody phospho-histone H3 (PHH3) specifically highlights MFs. Counting MFs on slides stained against PHH3 leads to higher agreement among raters and has therefore recently been used as a ground truth for the annotation of MFs in H&E. However, as PHH3 facilitates the recognition of cells indistinguishable from H&E stain alone, the use of this ground truth could potentially introduce noise into the H&E-related dataset, impacting model performance. This study analyzes the impact of PHH3-assisted MF annotation on inter-rater reliability and object level agreement through an extensive multi-rater experiment. We found that the annotators' object-level agreement increased when using PHH3-assisted labeling. Subsequently, MF detectors were evaluated on the resulting datasets to investigate the influence of PHH3-assisted labeling on the models' performance. Additionally, a novel dual-stain MF detector was developed to investigate the interpretation-shift of PHH3-assisted labels used in H&E, which clearly outperformed single-stain detectors. However, the PHH3-assisted labels did not have a positive effect on solely H&E-based models. The high performance of our dual-input detector reveals an information mismatch between the H&E and PHH3-stained images as the cause of this effect., Comment: 10 pages, 5 figures, 1 Table

Adversarial training aims to defend against *adversaries*: malicious opponents whose sole aim is to harm predictive performance in any way possible - a rather harsh perspective, which we assert results in unnecessarily conservative models. Instead, we propose to model opponents as simply pursuing their own goals, rather than working directly against the classifier. Employing tools from strategic modeling, our approach uses knowledge or beliefs regarding the opponent's possible incentives as inductive bias for learning. Our method of *strategic training* is designed to defend against opponents within an *incentive uncertainty set*: this resorts to adversarial learning when the set is maximal, but offers potential gains when it can be appropriately reduced. We conduct a series of experiments that show how even mild knowledge regarding the adversary's incentives can be useful, and that the degree of potential gains depends on how incentives relate to the structure of the learning task.

This study examined which variables predicted the sense of presence (being there) in an immersive audio experience, with a focus on the impacts of immersion technology (headphones with spatialised sound versus speaker with 2D stereo sound), the nature of the audio experience, the narrative content, participants emotional and cognitive engagement and personal characteristics such as age and gender. Museum visitors listened to a story on the One Story, Many Voices immersive audio installation. A convergent mixed-methods design was used, including multiple regression analysis of survey data (n 185) and relational analysis of interview data (n 9). This study found mixed methods support from both surveys and interviews to suggest that presence was predicted by the audio quality (especially the audio being perceived as different from other audio stories they had listened to in the past), cognitive engagement (especially being totally absorbed), the narrative (the story and how it was told) and some support was found for emotional engagement (especially feeling a connection with the storyteller). Effects of the environment and immersion technology on presence received only either quantitative or qualitative support. No influences of personal characteristics were found. Findings are relevant for academics, immersive experience (XR) commissioners and developers.

The recently introduced Consistency models pose an efficient alternative to diffusion algorithms, enabling rapid and good quality image synthesis. These methods overcome the slowness of diffusion models by directly mapping noise to data, while maintaining a (relatively) simpler training. Consistency models enable a fast one- or few-step generation, but they typically fall somewhat short in sample quality when compared to their diffusion origins. In this work we propose a novel and highly effective technique for post-processing Consistency-based generated images, enhancing their perceptual quality. Our approach utilizes a joint classifier-discriminator model, in which both portions are trained adversarially. While the classifier aims to grade an image based on its assignment to a designated class, the discriminator portion of the very same network leverages the softmax values to assess the proximity of the input image to the targeted data manifold, thereby serving as an Energy-based Model. By employing example-specific projected gradient iterations under the guidance of this joint machine, we refine synthesized images and achieve an improved FID scores on the ImageNet 64x64 dataset for both Consistency-Training and Consistency-Distillation techniques.

Image editing has advanced significantly with the introduction of text-conditioned diffusion models. Despite this progress, seamlessly adding objects to images based on textual instructions without requiring user-provided input masks remains a challenge. We address this by leveraging the insight that removing objects (Inpaint) is significantly simpler than its inverse process of adding them (Paint), attributed to the utilization of segmentation mask datasets alongside inpainting models that inpaint within these masks. Capitalizing on this realization, by implementing an automated and extensive pipeline, we curate a filtered large-scale image dataset containing pairs of images and their corresponding object-removed versions. Using these pairs, we train a diffusion model to inverse the inpainting process, effectively adding objects into images. Unlike other editing datasets, ours features natural target images instead of synthetic ones; moreover, it maintains consistency between source and target by construction. Additionally, we utilize a large Vision-Language Model to provide detailed descriptions of the removed objects and a Large Language Model to convert these descriptions into diverse, natural-language instructions. We show that the trained model surpasses existing ones both qualitatively and quantitatively, and release the large-scale dataset alongside the trained models for the community.

Even though novel imaging techniques have been successful in studying brain structure and function, the measured biological signals are often contaminated by multiple sources of noise, arising due to e.g. head movements of the individual being scanned, limited spatial/temporal resolution, or other issues specific to each imaging technology. Data preprocessing (e.g. denoising) is therefore critical. Preprocessing pipelines have become increasingly complex over the years, but also more flexible, and this flexibility can have a significant impact on the final results and conclusions of a given study. This large parameter space is often referred to as multiverse analyses. Here, we provide conceptual and practical tools for statistical analyses that can aggregate multiple pipeline results along with a new sensitivity analysis testing for hypotheses across pipelines such as "no effect across all pipelines" or "at least one pipeline with no effect". The proposed framework is generic and can be applied to any multiverse scenario, but we illustrate its use based on positron emission tomography data.

The QUILT-1M dataset is the first openly available dataset containing images harvested from various online sources. While it provides a huge data variety, the image quality and composition is highly heterogeneous, impacting its utility for text-conditional image synthesis. We propose an automatic pipeline that provides predictions of the most common impurities within the images, e.g., visibility of narrators, desktop environment and pathology software, or text within the image. Additionally, we propose to use semantic alignment filtering of the image-text pairs. Our findings demonstrate that by rigorously filtering the dataset, there is a substantial enhancement of image fidelity in text-to-image tasks., Comment: 4 pages (short paper)

It has been recently proposed that Hawking evaporation might slow down after a black hole has lost about half of its mass. Such an effect, called "memory burden", is parameterized as a suppression in the mass loss rate by negative powers $n$ of the black hole entropy and could considerably extend the lifetime of a black hole. We study the impact of memory burden on the Primordial Black Hole (PBH) reheating scenario. Modified PBH evaporation leads to a significantly longer PBH dominated stage. Requiring that PBHs evaporate prior enough to Big Bang Nucleosynthesis shrinks the allowed PBH mass range. Indeed, we find that for $n>2.5$ the PBH reheating scenario is not viable. The frequency of the Gravitational Waves (GWs) induced by PBH number density fluctuations is bound to be larger than about a Hz, while the amplitude of the GW spectrum is enhanced due to the longer PBH dominated phase. Interestingly, we show that, in some models, the slope of the induced GW spectrum might be sensitive to the modifications to Hawking evaporation, proving it may be possible to test the "memory burden" effect via induced GWs. Lastly, we argue that our results could also apply to general modifications of Hawking evaporation., Comment: 18 pages+appendices, 7 figures; v2: corrected minor typos, matches published version

In numerous studies, deep learning algorithms have proven their potential for the analysis of histopathology images, for example, for revealing the subtypes of tumors or the primary origin of metastases. These models require large datasets for training, which must be anonymized to prevent possible patient identity leaks. This study demonstrates that even relatively simple deep learning algorithms can re-identify patients in large histopathology datasets with substantial accuracy. We evaluated our algorithms on two TCIA datasets including lung squamous cell carcinoma (LSCC) and lung adenocarcinoma (LUAD). We also demonstrate the algorithm's performance on an in-house dataset of meningioma tissue. We predicted the source patient of a slide with F1 scores of 50.16 % and 52.30 % on the LSCC and LUAD datasets, respectively, and with 62.31 % on our meningioma dataset. Based on our findings, we formulated a risk assessment scheme to estimate the risk to the patient's privacy prior to publication., Comment: 20 pages, 7 figures, 2 tables

To some, the advent of artificial intelligence (AI) promises better decision-making and increased military effectiveness while reducing the influence of human error and emotions. However, there is still debate about how AI systems, especially large language models (LLMs) that can be applied to many tasks, behave compared to humans in high-stakes military decision-making scenarios with the potential for increased risks towards escalation. To test this potential and scrutinize the use of LLMs for such purposes, we use a new wargame experiment with 214 national security experts designed to examine crisis escalation in a fictional U.S.-China scenario and compare the behavior of human player teams to LLM-simulated team responses in separate simulations. Here, we find that the LLM-simulated responses can be more aggressive and significantly affected by changes in the scenario. We show a considerable high-level agreement in the LLM and human responses and significant quantitative and qualitative differences in individual actions and strategic tendencies. These differences depend on intrinsic biases in LLMs regarding the appropriate level of violence following strategic instructions, the choice of LLM, and whether the LLMs are tasked to decide for a team of players directly or first to simulate dialog between a team of players. When simulating the dialog, the discussions lack quality and maintain a farcical harmony. The LLM simulations cannot account for human player characteristics, showing no significant difference even for extreme traits, such as "pacifist" or "aggressive sociopath." When probing behavioral consistency across individual moves of the simulation, the tested LLMs deviated from each other but generally showed somewhat consistent behavior. Our results motivate policymakers to be cautious before granting autonomy or following AI-based strategy recommendations., Comment: Updated with new human participant results and added new LLM to results; fixed error in Table 1; all claims unaffected

Vision-Language (VL) models have gained significant research focus, enabling remarkable advances in multimodal reasoning. These architectures typically comprise a vision encoder, a Large Language Model (LLM), and a projection module that aligns visual features with the LLM's representation space. Despite their success, a critical limitation persists: the vision encoding process remains decoupled from user queries, often in the form of image-related questions. Consequently, the resulting visual features may not be optimally attuned to the query-specific elements of the image. To address this, we introduce QA-ViT, a Question Aware Vision Transformer approach for multimodal reasoning, which embeds question awareness directly within the vision encoder. This integration results in dynamic visual features focusing on relevant image aspects to the posed question. QA-ViT is model-agnostic and can be incorporated efficiently into any VL architecture. Extensive experiments demonstrate the effectiveness of applying our method to various multimodal architectures, leading to consistent improvement across diverse tasks and showcasing its potential for enhancing visual and scene-text understanding.

The increasing use of transformer-based large language models brings forward the challenge of processing long sequences. In document visual question answering (DocVQA), leading methods focus on the single-page setting, while documents can span hundreds of pages. We present GRAM, a method that seamlessly extends pre-trained single-page models to the multi-page setting, without requiring computationally-heavy pretraining. To do so, we leverage a single-page encoder for local page-level understanding, and enhance it with document-level designated layers and learnable tokens, facilitating the flow of information across pages for global reasoning. To enforce our model to utilize the newly introduced document tokens, we propose a tailored bias adaptation method. For additional computational savings during decoding, we introduce an optional compression stage using our compression-transformer (C-Former),reducing the encoded sequence length, thereby allowing a tradeoff between quality and latency. Extensive experiments showcase GRAM's state-of-the-art performance on the benchmarks for multi-page DocVQA, demonstrating the effectiveness of our approach.

Numerous prognostic factors are currently assessed histopathologically in biopsies of canine mast cell tumors to evaluate clinical behavior. In addition, PCR analysis of the c-Kit exon 11 mutational status is often performed to evaluate the potential success of a tyrosine kinase inhibitor therapy. This project aimed at training deep learning models (DLMs) to identify the c-Kit-11 mutational status of MCTs solely based on morphology without additional molecular analysis. HE slides of 195 mutated and 173 non-mutated tumors were stained consecutively in two different laboratories and scanned with three different slide scanners. This resulted in six different datasets (stain-scanner variations) of whole slide images. DLMs were trained with single and mixed datasets and their performances was assessed under scanner and staining domain shifts. The DLMs correctly classified HE slides according to their c-Kit 11 mutation status in, on average, 87% of cases for the best-suited stain-scanner variant. A relevant performance drop could be observed when the stain-scanner combination of the training and test dataset differed. Multi-variant datasets improved the average accuracy but did not reach the maximum accuracy of algorithms trained and tested on the same stain-scanner variant. In summary, DLM-assisted morphological examination of MCTs can predict c-Kit-exon 11 mutational status of MCTs with high accuracy. However, the recognition performance is impeded by a change of scanner or staining protocol. Larger data sets with higher numbers of scans originating from different laboratories and scanners may lead to more robust DLMs to identify c-Kit mutations in HE slides., Comment: 17 pages, 3 figures, 4 tables

We develop the recipe to separate the spectral counterparts of the AGN NGC 1275 from the emission of the Perseus cluster surrounding it in the spectra observed by Suzaku/XIS cameras with no usage of the spectral fitting models. The Perseus cluster emission reaches higher energies than is typical for the most AGN-situated dense surroundings (i.e. up to 9-10 keV). That is why the separation between the AGN and cluster spectra is especially important in this case. To avoid the degeneracy due to the huge quantity of the spectral fitting parameters such as abundances of elements the cluster consists of, thermal and Compton emission of the nucleus itself, and the jet SSC/IC emission spectral parameters as well we prefer to avoid the spectral fitting usage to perform this task. Instead, we use the spatial resolution of the components and double background subtracting. For this purpose we choose the following regions to collect all the photons from them: (1) circular or square-shaped region around the source (AGN); (2) ring-shaped (or non-overlapped square) region surrounding the AGN (for cluster); (3) remote empty circular region for the background. Having collected the photons from those regions we subtract the background (i.e. photons from the third region) from the source and cluster spectra. Next, we subtract the re-normalized cluster counts from the AGN spectrum; using the relation between the emission line amplitudes in the AGN and cluster spectra as the renormalization coefficient. We have performed this procedure on the whole set of the Suzaku/XIS observational data for NGC 1275 to obtain the cleaned spectra and light curve of the AGN emission in this system., Comment: 18 pages, 6 pages, 8 tables