A major challenge is thus in designing sample-efficient agents that can transfer their existing knowledge to solve new tasks quickly.

The ability to compose learned skills to solve new tasks is an important property for lifelong-learning agents.

Methods for probability updating, of which Bayesian conditionalization is the most well-known and widely used, are modeling tools that aim to represent the process of modifying an initial epistemic state, typically represented by a prior probability function P, which is adjusted in light of new information. Notably, updating methods and conditional sentences seem to intuitively share a deep connection, as is evident in the case of conditionalization. The present work contributes to this line of research and aims at shedding new light on the relationship between updating methods and conditional connectives. Departing from previous literature that often focused on a specific type of conditional or a particular updating method, our goal is to prove general results concerning the connection between conditionals and their probabilities. This will allow us to characterize the probabilities of certain conditional connectives and to understand what class of updating procedures can be represented using specific conditional connectives. Broadly, we adopt a general perspective that encompasses a large class of conditionals and a wide range of updating methods, enabling us to prove some general results concerning their interrelation.

We thank the reviewers for their constructive feedback and hope to clarify and address their concerns in this response. UVF As may help with more complex settings. We will add this explanation in the paper. Note that Assump 1 does not require binary rewards in terminal states (also see discussion after Assump 1). "stay", such that a goal position only becomes terminal if the agent chooses to stay in it.

The ability to compose learned skills to solve new tasks is an important property for lifelong-learning agents. This allows us to formulate new tasks in terms of the negation, disjunction and conjunction of a set of base tasks. We then show that by learning goal-oriented value functions and restricting the transition dynamics of the tasks, an agent can solve these new tasks with no further learning. We prove that by composing these value functions in specific ways, we immediately recover the optimal policies for all tasks expressible under the Boolean algebra. We verify our approach in two domains---including a high-dimensional video game environment requiring function approximation---where an agent first learns a set of base skills, and then composes them to solve a super-exponential number of new tasks.

We propose a novel topological perspective on data languages recognizable by orbit-finite nominal monoids. For this purpose, we introduce pro-orbit-finite nominal topological spaces. Assuming globally bounded support sizes, they coincide with nominal Stone spaces and are shown to be dually equivalent to a subcategory of nominal boolean algebras. Recognizable data languages are characterized as topologically clopen sets of pro-orbit-finite words. In addition, we explore the expressive power of pro-orbit-finite equations by establishing a nominal version of Reiterman's pseudovariety theorem.

We present a logical system that combines the well-known classical epistemic concepts of belief and knowledge with a concept of evidence such that the intuitive principle \textit{`evidence yields belief and knowledge'} is satisfied. Our approach relies on previous works of the first author \cite{lewjlc2, lewigpl, lewapal} who introduced a modal system containing $S5$-style principles for the reasoning about intutionistic truth (i.e. \textit{proof}) and, inspired by \cite{artpro}, combined that system with concepts of \textit{intuitionistic} belief and knowledge. We consider that combined system and replace the constructive concept of \textit{proof} with a classical notion of \textit{evidence}. This results in a logic that combines modal system $S5$ with classical epistemic principles where $\square\varphi$ reads as `$\varphi$ is evident' in an epistemic sense. Inspired by \cite{lewapal}, and in contrast to the usual possible worlds semantics found in the literature, we propose here a relational, frame-based semantics where belief and knowledge are not modeled via accessibility relations but directly as sets of propositions (sets of sets of worlds).

Contrary to popular belief, the history of machine learning, which enables machines to learn tasks for which they are not specifically programmed, and train themselves in unfamiliar environments, goes back to 17th century. Machine learning is a powerful tool for implementing artificial intelligence technologies. Because of its ability to learn and make decisions, machine learning is frequently referred to as AI, even though it is technically a subdivision of AI technology. Until the late 1970s, machine learning was only another component of AI's progress. It then diverged and evolved on its own, as machine learning has emerged as an important function in cloud computing and e-Commerce. ML is a vital enabler in many cutting-edge technology areas of our times. Scientists are currently working on Quantum Machine Learning approaches.