Cognitive warfare has emerged as a central feature of modern conflict, yet it remains inconsistently defined and difficult to evaluate. Existing approaches often treat cognitive operations as a subset of information operations, limiting the ability to assess cognitive attacker-defender interactions or determine when advantage has been achieved. This article proposes a unified definition of cognitive warfare, introduces an interaction framework grounded in the OODA loop, and identifies measurable attributes associated with cognitive superiority. To illustrate the use of the framework, a notional case study demonstrates how these concepts can be applied to assess cognitive attacks and defenses in a contested environment. Thus, the framework provides joint force leaders and analysts with a practical foundation for understanding, comparing, and evaluating cognitive warfare campaigns.
Extracting structured data from unstructured text using large language models (LLMs) becomes challenging when target schemas are large and complex. In such cases, including the full schema in the prompt increases cost and latency, risks lost-in-the-middle performance degradation, and can exceed context length limits. We propose SchemaRAG, a retrieval-augmented generation (RAG) framework that dynamically prunes the output schema space for schema-conditioned information extraction tasks by leveraging schema metadata and few-shot examples when available. We evaluate SchemaRAG on real-world healthcare and e-commerce datasets. Our results show that SchemaRAG can achieve up to an 8.8% increase in micro-F1, a 47% reduction in latency, and a 48% reduction in token costs, demonstrating its practicality for large-schema extraction.
Social media enables data-driven analysis of public opinion on contested issues. Target-Stance Extraction (TSE) is the task of identifying the target discussed in a document and the document's stance towards that target. Many works classify stance towards a given target in a multilingual setting, but all prior work in TSE is English-only. This work introduces the first multilingual TSE benchmark, spanning Catalan, Estonian, French, Italian, Mandarin, and Spanish corpora. It manages to extend the original TSE pipeline to a multilingual setting without requiring separate models for each language. Our model pipeline achieves a modest F1 score of 12.78, underscoring the increased difficulty of the multilingual task relative to English-only setups and highlighting target prediction as the primary bottleneck. We are also the first to demonstrate the sensitivity of TSE's F1 score to different target verbalizations. Together these serve as a much-needed baseline for resources, algorithms, and evaluation criteria in multilingual TSE.
Advances in deep learning have opened an era of abundant and accurate predicted protein structures; however, similar progress in protein ensembles has remained elusive. This review highlights several recent research directions towards AI-based predictions of protein ensembles, including coarse-grained force fields, generative models, multiple sequence alignment perturbation methods, and modeling of ensemble descriptors. An emphasis is placed on realistic assessments of the technological maturity of current methods, the strengths and weaknesses of broad families of techniques, and promising machine learning frameworks at an early stage of development. We advocate for "closing the loop" between model training, simulation, and inference to overcome challenges in training data availability and to enable the next generation of models.
Toxic language detection is crucial for creating safer online environments and limiting the spread of harmful content. While toxic language detection has been under-explored in Persian, the current work compares different methods for this task, including fine-tuning, data enrichment, zero-shot and few-shot learning, and cross-lingual transfer learning. What is especially compelling is the impact of cultural context on transfer learning for this task: We show that the language of a country with cultural similarities to Persian yields better results in transfer learning. Conversely, the improvement is lower when the language comes from a culturally distinct country. Warning: This paper contains examples of toxic language that may disturb some readers. These examples are included for the purpose of research on toxic detection.
In this paper we begin the study of well-failed graphs, that is, graphs in which every maximal failed zero forcing set is a maximum failed zero forcing set, or equivalently, in which every minimal fort is a minimum fort. We characterize trees that are well-failed. Along the way, we prove that the set of vertices in a graph that are not in any minimal fort is identical to the set of vertices that are in no minimal zero forcing set, which allows us to characterize vertices in a tree that are in no minimal fort.
Spectropolarimetry, the observation of polarization and intensity as a function of wavelength, is a powerful tool in stellar astrophysics. It is particularly useful for characterizing stars and circumstellar material, and for tracing the influence of magnetic fields on a host star and its environment. Maintaining modern, flexible, and accessible computational tools that enable spectropolarimetric studies is thus essential. The SpecpolFlow package is a new, completely Pythonic workflow for analyzing stellar spectropolarimetric observations. Its suite of tools provides a user-friendly interface for working with data from an assortment of instruments and telescopes. SpecpolFlow contains tools for spectral normalization and visualization, the extraction of Least-Squares Deconvolution (LSD) profiles, the generation and optimization of line masks for LSD analyses, and the calculation of longitudinal magnetic field measurements from the LSD profiles. It also provides Python classes for the manipulation of spectropolarimetric products. The SpecpolFlow website includes an array of tutorials that guide users through common analysis cases using the software. SpecpolFlow is distributed as a fr
Over the past decade, Rydberg atom electric field sensors have been under investigation as potential alternatives or complements to conventional antenna-based receivers for select applications in RF communications, remote sensing, and precision metrology. To understand the potential utility of these devices for various use cases, it is crucial to develop models that accurately predict key performance metrics such as instantaneous bandwidth and dynamic range. However, existing numerical models require solving a large set of coupled differential equations that is computationally intensive and lengthy to solve. We present an analytic approach that can be used to derive an impulse response function that allows up to two orders-of-magnitude reduction in computation time compared to the full time-dependent integration of the equations of motion. This approach can be used to enable rapid assessments of the Rydberg sensor's response to various waveforms.
Cyber cognitive attacks leverage disruptive innovations (DIs) to exploit psychological biases and manipulate decision-making processes. Emerging technologies, such as AI-driven disinformation and synthetic media, have accelerated the scale and sophistication of these threats. Prior studies primarily categorize current cognitive attack tactics, lacking predictive mechanisms to anticipate future DIs and their malicious use in cognitive attacks. This paper addresses these gaps by introducing a novel predictive methodology for forecasting the emergence of DIs and their malicious uses in cognitive attacks. We identify trends in adversarial tactics and propose proactive defense strategies.
As disinformation-driven cognitive attacks become increasingly sophisticated, the ability to quantify their impact is essential for advancing cybersecurity defense strategies. This paper presents a novel framework for measuring the engagement effectiveness of cognitive attacks by introducing a weighted interaction metric that accounts for both the type and volume of user engagement relative to the number of attacker-generated transmissions. Applying this model to real-world disinformation campaigns across social media platforms, we demonstrate how the metric captures not just reach but the behavioral depth of user engagement. Our findings provide new insights into the behavioral dynamics of cognitive warfare and offer actionable tools for researchers and practitioners seeking to assess and counter the spread of malicious influence online.
Implicit discourse relation recognition involves determining relationships that hold between spans of text that are not linked by an explicit discourse connective. In recent years, the pre-train, prompt, and predict paradigm has emerged as a promising approach for tackling this task. However, previous work solely relied on manual verbalizers for implicit discourse relation recognition, which suffer from issues of ambiguity and even incorrectness. To overcome these limitations, we leverage the prototypes that capture certain class-level semantic features and the hierarchical label structure for different classes as the verbalizer. We show that our method improves on competitive baselines. Besides, our proposed approach can be extended to enable zero-shot cross-lingual learning, facilitating the recognition of discourse relations in languages with scarce resources. These advancement validate the practicality and versatility of our approach in addressing the issues of implicit discourse relation recognition across different languages.
Doubly robust estimators have gained popularity in the field of causal inference due to their ability to provide consistent point estimates when either an outcome or exposure model is correctly specified. However, for nonrandomized exposures the influence function based variance estimator frequently used with doubly robust estimators of the average causal effect is only consistent when both working models (i.e., outcome and exposure models) are correctly specified. Here, the empirical sandwich variance estimator and the nonparametric bootstrap are demonstrated to be doubly robust variance estimators. That is, they are expected to provide valid estimates of the variance leading to nominal confidence interval coverage when only one working model is correctly specified. Simulation studies illustrate the properties of the influence function based, empirical sandwich, and nonparametric bootstrap variance estimators in the setting where parametric working models are assumed. Estimators are applied to data from the Improving Pregnancy Outcomes with Progesterone (IPOP) study to estimate the effect of maternal anemia on birth weight among women with HIV.
Recent work to enhance data partitioning strategies for more realistic model evaluation face challenges in providing a clear optimal choice. This study addresses these challenges, focusing on morphological segmentation and synthesizing limitations related to language diversity, adoption of multiple datasets and splits, and detailed model comparisons. Our study leverages data from 19 languages, including ten indigenous or endangered languages across 10 language families with diverse morphological systems (polysynthetic, fusional, and agglutinative) and different degrees of data availability. We conduct large-scale experimentation with varying sized combinations of training and evaluation sets as well as new test data. Our results show that, when faced with new test data: (1) models trained from random splits are able to achieve higher numerical scores; (2) model rankings derived from random splits tend to generalize more consistently.
The advent of social media has given rise to numerous ethical challenges, with hate speech among the most significant concerns. Researchers are attempting to tackle this problem by leveraging hate-speech detection and employing language models to automatically moderate content and promote civil discourse. Unfortunately, recent studies have revealed that hate-speech detection systems can be misled by adversarial attacks, raising concerns about their resilience. While previous research has separately addressed the robustness of these models under adversarial attacks and their interpretability, there has been no comprehensive study exploring their intersection. The novelty of our work lies in combining these two critical aspects, leveraging interpretability to identify potential vulnerabilities and enabling the design of targeted adversarial attacks. We present a comprehensive and comparative analysis of adversarial robustness exhibited by various hate-speech detection models. Our study evaluates the resilience of these models against adversarial attacks using explainability techniques. To gain insights into the models' decision-making processes, we employ the Local Interpretable Mode
Effective cyber threat recognition and prevention demand comprehensible forecasting systems, as prior approaches commonly offer limited and, ultimately, unconvincing information. We introduce Simplified Plaintext Language (SPLAIN), a natural language generator that converts warning data into user-friendly cyber threat explanations. SPLAIN is designed to generate clear, actionable outputs, incorporating hierarchically organized explanatory details about input data and system functionality. Given the inputs of individual sensor-induced forecasting signals and an overall warning from a fusion module, SPLAIN queries each signal for information on contributing sensors and data signals. This collected data is processed into a coherent English explanation, encompassing forecasting, sensing, and data elements for user review. SPLAIN's template-based approach ensures consistent warning structure and vocabulary. SPLAIN's hierarchical output structure allows each threat and its components to be expanded to reveal underlying explanations on demand. Our conclusions emphasize the need for designers to specify the "how" and "why" behind cyber warnings, advocate for simple structured templates in
In zero forcing, the focus is typically on finding the minimum cardinality of any zero forcing set in the graph; however, the number of cardinalities between $0$ and the number of vertices in the graph for which there are both zero forcing sets and sets that fail to be zero forcing sets is not well known. In this paper, we introduce the zero forcing span of a graph, which is the number of distinct cardinalities for which there are sets that are zero forcing sets and sets that are not. We introduce the span within the context of standard zero forcing and skew zero forcing as well as for standard zero forcing on directed graphs. We characterize graphs with high span and low span of each type, and also investigate graphs with special zero forcing polynomials.
While randomized controlled trials (RCTs) are critical for establishing the efficacy of new therapies, there are limitations regarding what comparisons can be made directly from trial data. RCTs are limited to a small number of comparator arms and often compare a new therapeutic to a standard of care which has already proven efficacious. It is sometimes of interest to estimate the efficacy of the new therapy relative to a treatment that was not evaluated in the same trial, such as a placebo or an alternative therapy that was evaluated in a different trial. Such multi-study comparisons are challenging because of potential differences between trial populations that can affect the outcome. In this paper, two bridging estimators are considered that allow for comparisons of treatments evaluated in different trials using data fusion methods to account for measured differences in trial populations. A "multi-span'' estimator leverages a shared arm between two trials, while a "single-span'' estimator does not require a shared arm. A diagnostic statistic that compares the outcome in the standardized shared arms is provided. The two estimators are compared in simulations, where both estimator
A graph in which all minimal zero forcing sets are in fact minimum size is called ``well-forced." This paper characterizes well-forced trees and presents an algorithm for determining which trees are well-forced. Additionally, we characterize which vertices in a tree are contained in no minimal zero forcing set.
Implicit discourse relation recognition is a challenging task that involves identifying the sense or senses that hold between two adjacent spans of text, in the absence of an explicit connective between them. In both PDTB-2 and PDTB-3, discourse relational senses are organized into a three-level hierarchy ranging from four broad top-level senses, to more specific senses below them. Most previous work on implicit discourse relation recognition have used the sense hierarchy simply to indicate what sense labels were available. Here we do more -- incorporating the sense hierarchy into the recognition process itself and using it to select the negative examples used in contrastive learning. With no additional effort, the approach achieves state-of-the-art performance on the task.