Abductive reasoning

About: Abductive reasoning is a research topic. Over the lifetime, 1917 publications have been published within this topic receiving 44645 citations. The topic is also known as: abduction & abductive inference.

Temporal Abductive Reasoning based Intelligent Alarm for Power System

Abstract: Abductive Reasoning is known as using domain theory to infer about reasons or explanations of observations and results. It is one basic inference mechanism for diagnosis problems. Abductive Reasoning based inference can find out causes and evolution processes, which are encouraging results for diagnosis. This paper develops the temporal relationship of faults and alarms in power grid, and raises a suit of abductive temporal reasoning algorithms to diagnose fault evolution process using observed alarms. The algorithms developed in this paper can recognize break alarms, exception alarms and missing alarms. The result of this paper can be taken as a function of intelligent alarm method for power system, and used with other methods for further fault diagnosis. The numerical test is given in the last part of paper.

The Extra-Theoretical Dimension of Discovery. Extracting Knowledge by Abduction

Abstract: Science is one of the most creative forms of human reasoning. The recent epistemological and cognitive studies concentrate on the concept of abduction as a means to originate and refine new ideas. Traditional cognitive science accounts concerning abduction aim at illustrating discovery and creativity processes in terms of theoretical and "internal" aspects, by means of computational simulations and/or abstract cognitive models. A neglected issue, worth of a deepest investigation inside artificial intelligence, is that "discovery" is often related to a complex cognitive task involving the use and the manipulation of external world. Concrete manipulations of external world is a fundamental passage in the process of knowledge extraction and hypotheses generation: by a process of manipulative abduction it is possible to build prostheses for human minds, by interactingwith external objects and representations in a constructive way, and so by creating implicit knowledge through doing. This kind of embodied and unexpressed knowledge holds a key role in the subsequent processes of scientific comprehension and discovery. This paper aims at illustratingthe close relationship between external representations and creative processes in scientific explorations and understanding of phenomena.

Abductive Reasoning as Self-Supervision for Common Sense Question Answering.

Abstract: Question answering has seen significant advances in recent times, especially with the introduction of increasingly bigger transformer-based models pre-trained on massive amounts of data. While achieving impressive results on many benchmarks, their performances appear to be proportional to the amount of training data available in the target domain. In this work, we explore the ability of current question-answering models to generalize - to both other domains as well as with restricted training data. We find that large amounts of training data are necessary, both for pre-training as well as fine-tuning to a task, for the models to perform well on the designated task. We introduce a novel abductive reasoning approach based on Grenander's Pattern Theory framework to provide self-supervised domain adaptation cues or "pseudo-labels," which can be used instead of expensive human annotations. The proposed self-supervised training regimen allows for effective domain adaptation without losing performance compared to fully supervised baselines. Extensive experiments on two publicly available benchmarks show the efficacy of the proposed approach. We show that neural networks models trained using self-labeled data can retain up to $75\%$ of the performance of models trained on large amounts of human-annotated training data.

Case-Based Reasoning

Abstract: Case-based reasoning represents a memory-based, data-driven estimation method. In other words, it is an estimation method in which estimates are based solely on the analysis of quantitative project data and in which the data need to be available at the time of estimation.

Efficient inference method for computing an optimal solution in predicate-logic hypothetical reasoning

Abstract: Hypothetical reasoning, which is one type of abductive reasoning, is an important framework in the development of advanced knowledge-based systems. One problem with hypothetical reasoning is its slow inference speed, which is due to its nonmonotonic inference nature. A fast hypothetical reasoning system with predicate Horn clause expressions has been developed to overcome this problem. However, when the constraints for hypotheses are not strong, the number of hypotheses to be synthetized becomes too large to calculate. The paper presents an efficient hypothetical reasoning method combining best-first search, beam search and branch-and-bound search strategies for computing the optimal solution, which is the most desirable solution in many cases. The effectiveness of this method is shown experimentally using fault-diagnosis problems in logic circuits.