Inaccurate Rules Research Articles

Safe Abductive Learning in the Presence of Inaccurate Rules

Integrating complementary strengths of raw data and logical rules to improve the learning generalization has been recently shown promising and effective, e.g., abductive learning is one generic framework that can learn the perception model from data and reason between rules simultaneously. However, the performance would be seriously decreased when inaccurate logical rules appear, which may be even worse than baselines using only raw data. Efforts on this issue are highly desired while remain to be limited. This paper proposes a simple and effective safe abductive learning method to alleviate the harm caused by inaccurate rules. Unlike the existing methods which directly use all rules without correctness checks, it utilizes them selectively by constructing a graphical model with an adaptive reasoning process to prevent performance hazards. Theoretically, we show that induction and abduction are mutually beneficial, and can be rigorously justified from a classical maximum likelihood estimation perspective. Experiments on diverse tasks show that our method can tolerate at least twice as many inaccurate rules as accurate ones and achieve highly competitive performance while other methods can't. Moreover, the proposal can refine inaccurate rules and works well in extended weakly supervised scenarios.

Brazilian Presidential Pronouncements in the Pandemic: Effectiveness in Crisis Communication and Rule Properties.

This work analyzed the six official statements of the president of Brazil that were broadcast on radio and television during the first 4 months of COVID-19 contamination in the country, regarding the efficacy in communicating the crisis and dimensions of rules. We observed a higher frequency of ineffective excerpts in the statements, especially in the categories “effective fear incitement” and “respect.” The categories “speed” and “expression of empathy” showed high efficacy. Additionally, there was a higher recurrence of implicit and inaccurate rules and rules opposing the recommendations of experts. These results indicate that the analyzed statements were ineffective in crisis communication and control of behaviors combating the pandemic in Brazil. The analysis of governmental practices by behavioral science can be useful in the planning of public policies.

The Effects of Feedback Accuracy and Trainer Rules on Performance during an Analogue Task

ABSTRACT Feedback is the most commonly used intervention in Organizational Behavior Management (OBM), and recent research has examined both the effects of inaccurate feedback on performance as well as rumors created due to inaccurate rules. To date, no studies have assessed the effects of providing differing information regarding feedback accuracy during the training of an analogue work task. The present study assessed the relative effects of two independent variables on performance during an analogue work task: the information regarding feedback accuracy provided during the training (i.e., a rule given during training describing feedback accuracy) and the actual level of task feedback accuracy to which participants were exposed during an analogue work task (i.e., one-third [accurate feedback divided by three], accurate, and tripled [accurate feedback multiplied by three]). The results of the study showed that participants exposed to accurate feedback during the analogue work task performed better than participants exposed to one-third or tripled feedback. Implications, limitations of the study, and directions for future research are discussed.

Evaluation of Lag Schedules and Rules on Persistent Response Variability With Preschoolers in a Group.

In the present study, experimenters evaluated the influence of lag schedules of reinforcement in combination with accurate and inaccurate (complete and incomplete) rules on the response variability of naming category items for typically developing preschoolers in a group format. Results showed that when lag schedules were introduced with 2 categories, response variability generalized to the third category. Furthermore, after participants experienced the lag schedule, variability persisted when the contingency no longer required variability. Participants continued to vary their responses unless the rule and contingency required them to repeat responses. We discuss potential clinical applications of using lag schedules in a group format and including rules during teaching, as well as directions for future research in this area.

An Analysis of Rule Deletion Scheme in XCS on Reinforcement Learning Problem

The XCS classifier system is an evolutionary rule-based learning technique powered by a Q-learning like learning mechanism. It employs a global deletion scheme to delete rules from all rules covering all state-action pairs. However, the optimality of this scheme remains unclear owing to the lack of intensive analysis. We here introduce two deletion schemes: 1) local deletion, which can be applied to a subset of rules covering each state (a match set), and 2) stronger local deletion, which can be applied to a more specific subset covering each state-action pair (an action set). The aim of this paper is to reveal how the above three deletion schemes affect the performance of XCS. Our analysis shows that the local deletion schemes promote the elimination of inaccurate rules compared with the global deletion scheme. However, the stronger local deletion scheme occasionally deletes a good rule. We further show that the two local deletion schemes greatly improve the performance of XCS on a set of noisy maze problems. Although the localization strength of the proposed deletion schemes may require consideration, they can be adequate for XCS rather than the original global deletion scheme.

Efeito de Regras Inacuradas e Monitoramento sobre Desempenhos em Programas de Reforços

The aim of this study was to investigate the effects of monitoring on response rates in the presence of inaccurate rules. Six undergraduate students were exposed to an Fixed Interval (FI) FI multiple reinforcement schedule and then to an Extinction (EXT) EXT multiple reinforcement schedule. The experimenter was present (i.e., monitoring) during only one of the multiple-schedule components. Before sessions, participants received the written, inaccurate rule press the button quickly. Sessions lasted for 10 minutes. Each participant emitted high response rates that were non-differential between components with or without monitoring. A discussion in terms of a selectionist approach is carried out.

Boosting: foundations and algorithms

Boosting is an approach to machine learning based on the idea of creating a highly accurate predictor by combining many weak and inaccurate rules of thumb. A remarkably rich theory has evolved around boosting, with connections to a range of topics, including statistics, game theory, convex optimization, and information geometry. Boosting algorithms have also enjoyed practical success in such fields as biology, vision, and speech processing. At various times in its history, boosting has been perceived as mysterious, controversial, even paradoxical.This book, written by the inventors of the method, brings together, organizes, simplifies, and substantially extends two decades of research on boosting, presenting both theory and applications in a way that is accessible to readers from diverse backgrounds while also providing an authoritative reference for advanced researchers. With its introductory treatment of all material and its inclusion of exercises in every chapter, the book is appropriate for course use as well. The book begins with a general introduction to machine learning algorithms and their analysis; then explores the core theory of boosting, especially its ability to generalize; examines some of the myriad other theoretical viewpoints that help to explain and understand boosting; provides practical extensions of boosting for more complex learning problems; and finally presents a number of advanced theoretical topics. Numerous applications and practical illustrations are offered throughout.

Impact of Rule Accuracy on Productivity and Rumor in an Organizational Analog

This study examined the effects of inaccurate rules on generation of rumor among participants and their productivity in an organizational analog setting. Dyads of participants were given an explicit rule that described a contingency (i.e., number of points earned for correct responses). Experiments 1 and 2 utilized quasirandom and counterbalanced alternating treatments designs, respectively, in which conditions alternated such that participants were exposed to the rule-stated contingency (accurate condition) or a variation of the contingency that was stated in the rule (i.e., different number of points awarded, inaccurate conditions). Results indicate that exposure to inaccurate rules increases frequency and duration of rumor behavior. Participants exposed to alternating rules and only inaccurate rules exhibit lower correct responding on work tasks relative to those exposed to only accurate rules.

Integrating Academic Skills into First Year Curricula: Using Wood v. Lucy, Lady Duff-Gordon to Teach the Role of Facts in Legal Reasoning

Law professors of first year students often lament the quality of the work their students produce on essay exams, which are plagued with missed issues, incomplete or inaccurate rules, poor use of facts, and a misunderstanding of basic legal concepts and structures. This Article attributes this disconnect to the critical failing of legal education - the tendency of first-year law courses to focus exclusively on teaching doctrine at the expense of legal skills and reasoning. This Article discusses the need to reorient students’ thinking about learning the law, highlighting the critical importance of explicitly teaching academic skills across the doctrinal curriculum, rather than solely in skills-oriented classes. Developing a foundation of academic skills in the first year will prepare students to focus on the denser doctrines and inclusion of more practice-oriented skills required in their second and third years. There are myriad ways in which real legal methods objectives can be directly incorporated into a traditional first-year syllabus. The Article demonstrates how the popular contracts case Wood v. Lucy, Lady Duff-Gordon can be used as a vehicle to explicitly teach the skills of fact identification and fact analysis in legal reasoning. This case deals with the amorphous and highly fact-sensitive doctrines of good faith and best efforts and is especially good for teaching fact analysis because of the subtle, but artful, use of facts by Judge Benjamin Cardozo.

Subject and Author Index

Boosting is an approach to machine learning based on the idea of creating a highly accurate predictor by combining many weak and inaccurate rules of thumb. A remarkably rich theory has evolved around boosting, with connections to a range of topics, including statistics, game theory, convex optimization, and information geometry. Boosting algorithms have also enjoyed practical success in such fields as biology, vision, and speech processing. At various times in its history, boosting has been perceived as mysterious, controversial, even paradoxical.This book, written by the inventors of the method, brings together, organizes, simplifies, and substantially extends two decades of research on boosting, presenting both theory and applications in a way that is accessible to readers from diverse backgrounds while also providing an authoritative reference for advanced researchers. With its introductory treatment of all material and its inclusion of exercises in every chapter, the book is appropriate for course use as well. The book begins with a general introduction to machine learning algorithms and their analysis; then explores the core theory of boosting, especially its ability to generalize; examines some of the myriad other theoretical viewpoints that help to explain and understand boosting; provides practical extensions of boosting for more complex learning problems; and finally presents a number of advanced theoretical topics. Numerous applications and practical illustrations are offered throughout.

Manipulating the Illusion of Control: Variations in Gambling as a Function of Perceived Control Over Chance Outcomes

The present study examined the ability to experimentally manipulate the cognitive heuristic entitled the “illusion of control.” Five adult human females gambled at roulette for the opportunity to earn course extra-credit points. An alternating treatments design was utilized whereby in one condition subjects were allowed to select the location on the roulette board they placed their bets, and in the other condition subjects had to give their chips to the experimenter to select the location. In addition, after subjects had played the game for a number of trials, inaccurate rules related to this “illusion of control” were introduced in a multiple baseline fashion across subjects, later followed by accurate rules. Results show that the control heuristic may exist for subjects, yet that it can be brought under experimental control through the use of experimenter delivered instructions. Implications for a behavioral treatment of gambling and for a within-subject approach to the study of rule-governed behavior are discussed.

EFFECTS OF CONSEQUENCES OF ADVICE ON PATTERNS OF RULE CONTROL AND RULE CHOICE

Rules in the form of advice can inaccurately state the effects of recommended responses by overstating or understating size of the consequences. Three experiments investigated the effects of such inaccuracies on patterns of rule control and rule choice with female college students. In Experiment 1, signaled accurate, overstated, or understated rules specified that a given number of points would be earned by pressing a designated key. For some subjects, rules specified a number of points to be gained; for other subjects, rules specified a number of points to be lost from an amount given earlier. Point totals stated in the inaccurate rules averaged 25% more (overstated) or 25% less (understated) than those received. When subjects could choose either the response specified in the rule or an alternative response that produced an unpredictable number of points, they showed greater sensitivity to the inaccuracy of overstated rules than understated rules. In trials at the end of the experiment in which subjects could choose which rule to see, subjects did not always choose accurate rules and often chose inaccurate rules for which they had shown less sensitivity earlier. Experiment 2 replicated this pattern in which subjects could choose which type of rule to see on a greater number of trials. Some evidence suggested that subjects prefer an improvement from the outcomes promised to those later received. In Experiment 3, rules misstated by averages of 25% and 50% were compared. Evidence suggested that increasing the size of the misstatement reduced the discrimination of inaccurate rules from accurate ones.

Inaccurate experiments and environment-induced superselection rules

It is argued that approximate superselection rules induced by the environment cannot account for the emergence of definite measurement results in single experiments. The reason for this is that the inaccuracy necessary for an experiment failing to distinguish between exact and approximate mixtures requires the pointer observable to be strictly classical. This is shown in the case that the observables of physical systems generateW*-algebras.

Effects of changing contingencies on the behavior of depressed and nondepressed individuals.

The effects of rules versus shaping on the behavior of depressed and nondepressed individuals were compared. Extending the findings in the depressive realism literature to a learning paradigm, the behavior of depressed individuals was more sensitive to changing contingencies than was the behavior of nondepressed individuals. Contrary to hypotheses, however, this effect appeared due primarily to the nondepressive Ss' strategy of continuing to follow an experimenter's inaccurate rules. Results suggest the relative absence of self-presentational concerns may lead depressed individuals to be more accurate in judging environmental contingencies.

Effects of changing contingencies on the behavior of depressed and nondepressed individuals.

The effects of rules versus shaping on the behavior of depressed and nondepressed individuals were compared. Extending the findings in the depressive realism literature to a learning paradigm, the behavior of depressed individuals was more sensitive to changing contingencies than was the behavior of nondepressed individuals. Contrary to hypotheses, however, this effect appeared due primarily to the nondepressive Ss' strategy of continuing to follow an experimenter's inaccurate rules. Results suggest the relative absence of self-presentational concerns may lead depressed individuals to be more accurate in judging environmental contingencies.