Abstract In this paper we present the probabilistic typed natural deduction calculus TPTND, designed to reason about and derive trustworthiness properties of computational processes, like those underlying current AI applications. Derivability in TPTND is interpreted as process extracting $n$ samples possibly complex outputs with a certain frequency from given categorical distribution. We formalize trust for such form hypothesis testing on distance between intended probability. The main...

We put forward a proposal for logics handling both non-monotonic reasoning and boundedly rational agents. propose hierarchy of depth-bounded logics. As the parameter k, which controls level depth, increases, non-monotonicity decreases while inferential power, in classical sense, increases. Ultimately, these aim to provide approximations logic. present arguments evidence supporting adoption this framework.

Abstract A robot intended to monitor human behavior must account for the user’s reactions minimize his/her perceived discomfort. The possibility of learning user interaction preferences and changing robot’s accordingly may positively impact quality with robot. should approach without causing any discomfort or interference. In this work, we contribute implement a novel Reinforcement Learning (RL) navigation toward user. Our implementation is proof-of-concept that uses data gathered from...

In this paper we advocate the use of Inductive Logic Programming as a device for explaining black-box models, e.g. Support Vector Machines (SVMs), when they are used to learn user preferences. We present case study where ILP system ILASP explain output SVM classifiers trained on preference datasets. Explanations produced in terms weak constraints, which can be easily understood by humans. both global and local approximator SVMs, score its fidelity, discuss how prove useful interactive...

Some aspects of eLearning experience can be enhanced in a very natural way by using the basic tools offered fuzzy logic.As matter example, consider uncontrolled growth information produced collaborative-oriented context, which each participant (e.g.students, teachers) is able to insert and share new contents (e.g.comments, texts) concerning university course.All incrementally added pieces evaluated several ways: intervention "dictator" (e.g. teacher), rating form, or even according frequency...

In this contribution we begin to discuss the thesis that an analysis of similarities and differences typical methodologies human sciences, technology hard sciences show some unforeseen but strong between technologies. context fuzzy sets ideas provide useful tools which help render more quantitative without loosing connection with a purely descriptive analysis. These kinds considerations would have been hardly conceivable in setting XIX Century conception science. It is development...

Abstract We present and discuss a runtime architecture that integrates sensorial data classifiers with logic-based decision-making system in the context of an e-Health for rehabilitation children neuromotor disorders. In this application, perform task form games. The main aim is to derive set parameters child’s current level cognitive behavioral performance (e.g., engagement, attention, accuracy) from available sensors eye trackers, motion sensors, emotion recognition techniques) take...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

The simple act of perceiving, planning an action, predicting others’ intentions, and imagining the events a day in future are cognitive processes with something common, at least according to Active Inference Hypothesis: They all based on model constructed basis our inner structures as well past experiences, themselves formed perceptions internal external stimuli. As conceptual tool, that hypothesis can be traced back Kant’s metaphysical assumption about way which we form from sensory data...

We describe a general procedure for translating Epistemic Probabilistic Event Calculus (EPEC) action language domains into Answer Set Programs (ASP), and show how the Python-driven features of ASP solver Clingo can be used to provide efficient computation in this probabilistic setting. EPEC supports probabilistic, epistemic reasoning containing narratives that include both an agent's own executions environmentally triggered events. Some actions may belief-conditioned, some imperfect sensing...

In many different social contexts, communication allows a collective intelligence to emerge.However, correct way of exchanging information usually requires determined topological configurations the agents involved in process.Such configuration should take into account several parameters, e.g.agents positioning, their proximity and time efficiency communication.Our aim is present an algorithm, based on evolutionary programming, which optimizes placement arbitrarily shaped areas.In order show...

In this paper we present the probabilistic typed natural deduction calculus TPTND, designed to reason about and derive trustworthiness properties of computational processes, like those underlying current AI applications. Derivability in TPTND is interpreted as process extracting $n$ samples possibly complex outputs with a certain frequency from given categorical distribution. We formalize trust for such form hypothesis testing on distance between intended probability. The main advantage...

Argumentation is a very active research field of Artificial Intelligence concerned with the representation and evaluation arguments used in dialogues between humans and/or artificial agents. Acceptability semantics formal argumentation systems define criteria for acceptance or rejection arguments. Several software systems, known as solvers, have been developed to compute accepted/rejected using such criteria. These include that learn identify accepted non-interpretable methods. In this paper...