Well-formedness properties such as convexity of a Boolean formula can have impacts on the utility of the systems they express. For example, in the domain of access control policies, where a Boolean formula can capture the semantics of a policy (i.e., what it allows and what it denies), a convex formula would mean that the policy rules can be expressed in a favorable format that is otherwise not guaranteed. In this paper, we explore the problem of testing convexity of a disjunctive normal form formula where every term has exactly two literals in it (2DNF). In particular, we propose polynomial-time algorithms by leveraging a graph representation in which each term of the formula corresponds to an edge in the graph, allowing us to reduce the convexity check to problems of checking structural properties of graphs. We further classify these formulae into structural cases that guide the design of these algorithms. In particular, we obtain a linear-time algorithm for all-mixed 2DNF formulae, based on the analysis of the component graph of the term representation. An all-mixed 2DNF is one in which every term contains an unnegated and a negated literal. We then consider a strictly broader class of formulae that contains both all-mixed terms and all-positive (unnegated variables) terms. For this class, we develop a quadratic-time algorithm that analyzes a restricted subgraph of the component mixed-graph.
Wildfire smoke is increasingly degrading U.S. air quality via the emission and transport of pollutants. Smoke's direct role as a pollutant is well-documented; however, smoke also affects pollutant concentration indirectly by changing the shortwave actinic flux necessary for photochemical reactions. We compute smoke-driven changes in surface-level and boundary-layer downwelling actinic flux (F down arrow) at 550 and 380 nm (NO2 photolysis peak) along a 2018 Western wildfire Experiment for Cloud chemistry, Aerosol absorption, and Nitrogen (WE-CAN) research flight through the California Central Valley. The onboard HIAPER Airborne Radiation Package (HARP)-Actinic Flux instrument measured F down arrow. To assess changes in F down arrow relative to smoke-free conditions and at altitudes not sampled by the aircraft, we calculate F down arrow under assumed background and observed smoke conditions using the U.S. National Science Foundation (NSF) National Center for Atmospheric Research (NCAR) Tropospheric Ultraviolet and Visible (TUV) radiation model. Under smoke-impacted conditions, modeled F down arrow minorly underestimates HARP observations; the average modeled-to-measured ratio is 0.93 at 550 nm and 0.89 at 380 nm. Relative to modeled background conditions, observed (modeled) smoke-impacted F down arrow at 380 nm decreased by 24% (38%), 15% (24%), and 8% (18%) at 0-0.5 km, 0.5-1 km, and 1-1.5 km, respectively. At the ground, smoke decreased modeled F down arrow at 380 nm by 43%-likely an upper bound, as the modeled values slightly underestimate observations. As climate change drives more severe wildfire seasons, understanding smoke's impact on actinic flux is essential for constraining future air quality, and recent extreme seasons like 2018 offer opportunities for such analyses.
Although many perceive positive stereotypes as complimentary, research has revealed a variety of ways these stereotypes are harmful to oppressed groups, including their role in sustaining inequality and their operation as microaggressions. Despite these findings, a recent study demonstrated that liberal White Americans endorsed positive stereotypes of Black and Native Americans, which was partly due to their internal motivation to respond without prejudice. However, no prior work has explored liberal dominant group members' reasoning for their beliefs about positive stereotypes. To explore the apparent contradiction between internal motivation to respond without prejudice and endorsement of positive stereotypes, we conducted a study in which 698 self-identified liberal White Americans, the vast majority of whom were internally motivated to respond without prejudice, wrote reasons for their attitudes toward positive stereotypes of Black and Native Americans. Through qualitative thematic analysis, we identified a range of patterns in participants' written reasoning. Most participants endorsed positive stereotypes, and believed them helpful to the target groups, because they perceived them as complimentary. A smaller number were critical of positive stereotypes, most often because they believed stereotypes ignore differences among target group members. Few participants expressed concerns about how these stereotypes negatively impact interpersonal interaction and legitimate systemic oppression. Our findings suggest that there may be a paradox among some potential allies: they may reinforce oppression through beliefs they perceive to be helpful. We discuss theoretical implications of these findings, as well as practical implications for social change.
Emotion recognition has become a cornerstone of human-computer interaction (HCI), as it enables intelligent systems to respond dynamically to users’ affective states. Among the available modalities, electroencephalography (EEG) offers a non-invasive and reliable way of capturing neural activity associated with emotions. This work investigates EEG-based emotion classification using the MAHNOB-HCI dataset and applies a complete pipeline comprising signal preprocessing, feature extraction, and machine learning-based classification. Statistical, spectral, and wavelet-derived features were extracted, while ensemble classifiers such as Random Forest (RF) and XGBoost were employed for prediction. Different preprocessing strategies, including baseline correction and normalization, were examined to evaluate their influence on classification accuracy. Experimental results demonstrate that XGBoost consistently surpasses RF, particularly in recognizing valence and arousal states, with wavelet and power spectral density features contributing most to performance. The study underscores the effectiveness of EEG-driven emotion recognition and provides practical insights for affective computing and neurotechnology applications. Future research directions include improving subject-independent generalization and enhancing model robustness through the integration of larger datasets and deep learning approaches.
This article presents a collaborative speculative writing exercise in which international educators and researchers imagined the future of Generative Artificial Intelligence (GenAI) in education through three fictional scenarios. The exercise employed a relay structure: one participant began each story, and a different participant completed it. After the stories were finished, all participants read the scenarios they had not written and responded to a set of reflective questions. We treat the resulting scenarios and reflections as a qualitative, multi-voiced dataset and read them through a light analytical framework centred on postdigital temporality, relay discontinuity and recurring thematic tensions. The scenarios explore a GenAI-free Reset Day in a corporate training centre, a family navigating educational choices with the help of a GenAI assistant, and an underground school resisting a fully automated world. Situated within the postdigital condition, the article engages education fiction as a mode of collective inquiry that enacts the entanglement of human and technological agency it seeks to explore. The relay handoff introduced productive friction, and the reflections surfaced recurring questions about dependency, listening, measurement and the difficulty of translating situated, story-based imaginaries into policy recommendations. The contribution is both methodological, in theorising relay-based education fiction writing, and analytical, in showing how familiar postdigital concerns are reassembled through collaborative fictional handoffs.