The paper attempts to give a formal framework to capture the entire process of scientific discovery including hypothesis formation, reasoning, identifying contradictions, peer reviewing, reformulating and so on. Data mining can be seen as one step in this complex process of interactive learning of an empirical theory This paper uses the terminology from paraconsistent logic and paracomplete logic that extends Aristotle square in a hypercube of oppositions which defines or substantiates any step of the discovery process. The central formal notions are validated on a mathematical scientific discovery game, and an industrial application in the field of Drug Discovery illustrates how the presented framework combines different learning processes to predict pharmaco-kinetic properties (ADME-T) and adverse side effects of therapeutic drug molecules.
We start from the geometrical-logical extension of Aristotle’s square in [6,15] and [14], and study them from both syntactic and semantic points of view. Recall that Aristotle’s square under its modal form has the following four vertices: A is □α, E is \(\square\neg\alpha\), I is \(\neg\square\neg\alpha\) and O is \(\square\neg\alpha\), where α is a logical formula and □ is a modality which can be defined axiomatically within a particular logic known as S5 (classical or intuitionistic, depending on whether \(\neg\) is involutive or not) modal logic.
Since its inception, the field of machine learning has seen the advent of several learning paradigms, designed to frame the issues central to the learning activity, provide effective learning methods, and investigate the power and limitations inherent to the process of successful learning. In this article, we propose a formalization that underlies the key concepts of many such paradigms and discuss their relevance to scientific discovery, with the aim of assessing what scientists can expect from machines designed to assist them in their quest for the discovery of valid laws. We illustrate the formalization on several variations of a card game, and highlight the differences that paradigms impose on learners, as well as the assumptions they make on the nature of the learning process. We then use the formalization to describe a multi-agent interaction protocol, that has been inspired by these paradigms and that has been validated recently on some groups of agents. Finally, we propose extensions to this protocol.
In this paper, we describe a formal logical framework which we claim as essential to prove and to revise a model produced by combined ILP techniques. The dynamic process of proof embrace the supervision of the learning machine by a human, and this framework places the interpretation of contradictions in the heart of the interactive process which leads to a model which can be discussed, justified, and proven. We illustrate and validate this framework on an industrial application in the field of Drug Discovery, combining different learning processes to predict pharmaco-kinetic properties (ADME-T) and adverse side effects of therapeutic drug molecules.
This article studies machine learning paradigms from the point of view of human cognition. Indeed, conceptions in both machine learning and human learning evolved from a passive to an active conception of learning. Our objective is to provide an interaction protocol suited to both humans and machines to enable assisting human discoveries by learning machines. We identify the limitations of common machine learning paradigms in the context of scientific discovery, and we propose an extension inspired by game theory and multiagent systems. We present individual cognitive aspects of this protocol as well as social considerations, and we relate encouraging results concerning a game implementing it.
As mentioned in French secondary school official texts, teaching science implies teaching scientific process. This poses the problem of how to teach epistemology, as traditional science teaching is mostly dogmatic and based on contents. Previous studies show that pupils, science students and teachers mostly own positivist and realist spontaneous conceptions of science and scientific discovery. Here, we present the evaluation of the didactic impact of a network game, Eleusis+Nobel, on third year biology students who aim at becoming teachers. This cards game, based on a Popperian epistemology, has been designed to reproduce the scientific discovery process in a community. In the limits of our study, results obtained with classical social psychology tools indicate that students who played this game specifically assimilated the subjective dimension of knowledge and the role of the community in their conception of science, on the contrary to negative control students, who did not play.
The paper attempts to give a formal framework to capture the entire process of scientific discovery including hypothesis formation, reasoning, identifying contradictions, peer reviewing, reformulating and so on. Data mining can be seen as one step in this complex process of interactive learning of an empirical theory This paper uses the terminology from paraconsistent logic and paracomplete logic that extends Aristotle square in a hypercube of oppositions which defines or substantiates any step of the discovery process. The central formal notions are validated on a mathematical scientific discovery game, and an industrial application in the field of Drug Discovery illustrates how the presented framework combines different learning processes to predict pharmaco-kinetic properties (ADME-T) and adverse side effects of therapeutic drug molecules.
To the best of our knowledge, this paper is the first attempt to formalise a pragmatic logic of scientific discovery in a manner such that it can be realised by scientists assisted by machines. Using Institution Agents, we define a dialectic process to manage contradiction. This allows autoepistemic Institution Agents to learn from a supervised teaching process. We present an industrial application in the field of Drug Discovery, applying our system in the prediction of pharmaco-kinetic properties (ADME-T) and adverse side effects of therapeutic drug molecules.
This paper is an attempt to design an interaction protocol for a multi-agent learning platform to assist a human community in their task of scientific discovery. Designing tools to assist Scientific Discovery offers a challenging problematic, since the problems studied by scientists are not yet solved, and valid models are not yet available. It is therefore impossible to create a problem solver to simulate a given phenomenon and explain or predict facts. We propose to assist scientists with learning machines considered as adaptive problem solvers, to build interactively a consistent model suited for reasoning, simulating, predicting, and explaining facts. The interaction protocol presented in this paper is based on Angluin's “Learning from Different Teachers” [1] and we extend the original protocol to make it operational to assist scientists solve open problems. The main problem we deal with is that this learning model supposes the existence of teachers having previously solved the problem. These teachers are able to answer the learner's queries whereas this is not the case in the context of Scientific Discovery in which it is only possible to refute a model by finding experimental processes revealing contradictions. Our first contribution is to directly use Angluin's interaction protocol to let a machine learn a program that approximates the theory of a scientist, and to help him improve this theory. Our second contribution is to attenuate Angluin's protocol to take into account a social cognition level during which multiple scientists interact with each other by the means of publications and refutations of rival theories. The program learned by the machine can be included in a publication to avoid false refutations coming from a wrong interpretation of the theory.
We present our use of Aristotle's square and its extensions (see [Beziau03], [Moretti04]) to formalize a pragmatic logic of scientific discovery. We provide an interpretation of the resulting hypercubic structure to frame the paraconsistent and paracomplete representation of the computation of a theoretical predictive model corroborated by experimentation. This activity of producing a predictive and explicative model is at the core of the scientific interactive process of publication, refutation, and model confrontation that occurs during the construction of a consensual theory by a community. Finally, we describe in a constructive way how this logic can be achieved by a hierarchical and modular community of auto-epistemic and adaptive agents, as in [Sallantin06].
Old and recent theories stress that any understanding of the processes by which humans can learn requires to fully appreciate the relationships between the “nature of learning” and the “learning of nature.” From a constructivist viewpoint, acquiring knowledge is, like any human activity, dissociable neither from its underlying project nor from the knowing subject. The authors of this chapter relate the lessons from philosophy, psychology, didactics and ethics to their work in computational scientific discovery that aims at empowering learning machines with the task of assisting human researchers (Dartnell, Martin, Hagège, & Sallantin, 2008). The chapter concludes with didactical and ethical considerations.
