This article analyses how distinct argumentation spaces-media, law, and science-interact and collide, using Liebeck v. McDonald's as a legal case study and the hydroxychloroquine COVID-19 debate as a case of scientific dispute. Building on these two case studies, this analysis shows that arguments that intervene in many public debates originate from and move across different spaces, which employ different criteria, rules of evidence, and modes of reasoning for evaluating arguments. Furthermore, our study shows media-driven simplification, reframing, and truncation of nuanced expert exchanges, producing polarized public perceptions and policy consequences. The key features we identified include sliding arguments, repetition-driven strength (clones), shifting proof standards, hedgehog arguments, and zombie information, with implications for formalizing real-world argumentative dynamics and improving cross-space dialogue.
We introduce and develop the theory of Kalman-Galois connections on pseudocomplemented distributive lattices (KGp-algebras). We establish their fundamental properties, study KGp-Boolean filters and KGp-Boolean congruences, and prove a Glivenko-type theorem that relates KGp-algebras to Boolean gebras with additional operators. Using the Kalman construction, we prove categorical equivalence between the algebraic category of KGp-algebras and algebraic category of centered Kleene algebras with intuitionistic negation dowed with a unary operator g (gKANc-algebras). We also extend the Kalman construction to the Sendlewski construction and establish a categorical equivalence between the category of pairs consisting of a KGp-algebra and a Boolean KGp-filter and the algebraic category of gKAN-algebras. Additionally, we troduce the Monteiro construction, which allows us to associate to every gKANalgebra a gKANc-algebra into which it is embedded. These results enable us clarify the relationship between the Kalman and Monteiro constructions in case of gKAN-algebras.
Contradiction separation (CS) and its first-order version S-CS are multi-clause inference schemes for clausal refutation. They isolate a standard contradiction core within a clause set and derive a propagated clause from the remaining literals. This paper develops structural characterizations of non-unit standard contradictions that make such cores explicit and easier to identify. In propositional logic, we introduce a canonical dual-line family and prove that every instance is a standard contradiction. We study admissible literal extensions, define ladder structures as maximal dual-line extensions, and present a regular triple-line family with constructive generation schemes. We also analyze how dual-line cores compose via clause connections and give sufficient conditions under which the composed clause set remains a standard contradiction. In first-order logic, we exhibit clause families that are not standard contradictions syntactically but become standard contradictions after suitable instantiation and controlled clause reuse.
In precision animal husbandry, behavioral monitoring collars have become widely used tools for individualized and real-time herd management. These devices allow for the collection of individual animal activity data to be analyzed to identify behavioral patterns or specific events of interest. One of the most relevant events that can be detected by analyzing behavioral changes is estrus. In this context, the present paper analyzes and compares the performance of several supervised classification techniques combined with feature engineering for early estrus detection in dairy cows from intensive farms using behavioral data obtained from a commercial smart collar. For this purpose, twelve dairy cows were monitored in a Galician intensive farm, and customized models were developed for each animal. Several configurations of both the models and the input data used were evaluated. The results obtained were excellent, reaching, in some cases, F1-Score values of up to 1 in specific configurations and techniques. In addition, this research demonstrates a high variability in model performance between cows, highlighting the need to develop individualized animal models. It is also concluded that to obtain good results, it is essential to provide the model with a temporal context that includes the animal's previous behavior.
This article focuses on optimizing machine learning when each solution evaluation is costly and/or time-consuming. In particular, current deep learning models, which are complex and computationally intensive, rely on a large number of hyperparameters that need to be optimized. The purpose is to maximize the performance of these models by optimizing their hyperparameters and saving computational resources. In this paper, an informed search using Bayesian optimization is applied to tune the hyperparameters of the Convolutional Neural Network (CNN). Firstly, we analyze the performance of the Stochastic Gradient Descent with Momentum (SGDM) optimizer to learn the weights of CNN faster and more accurately. Then, we show the importance of the momentum hyperparameter for CNN learning. Finally, Bayesian optimization is applied to model the generalization performance of CNN as a sample from a Gaussian process (GP). The posterior distribution of the surrogate model is updated sequentially on the basis of a few evaluations of the true expensive function (CNN performance). The results confirm that CNN based on Bayesian optimization has promising performance for hyperparameter optimization when evaluations are expensive or time-limited. We also provide our script code at https://github.com/AmgadMonir/BO-Based-CNN.
