Reinforcement learning has shown great potential to address long-horizon tasks. However, most existing work lacks the ability to reason about functional parts of objects and extract the task semantics to facilitate ro...
详细信息
We automate deep step-by step reasoning in an LLM dialog thread by recursively exploring alternatives (OR-nodes) and expanding details (AND-nodes) up to a given depth. Starting from a single succinct task-specific ini...
详细信息
ISBN:
(纸本)9789819722990;9789819723003
We automate deep step-by step reasoning in an LLM dialog thread by recursively exploring alternatives (OR-nodes) and expanding details (AND-nodes) up to a given depth. Starting from a single succinct task-specific initiator we steer the automated dialog thread to stay focussed on the task by synthesizing a prompt that summarizes the depth-first steps taken so far. Our algorithm is derived from a simple recursive descent implementation of a Horn Clause interpreter, except that we accommodate our logic engine to fit the natural language reasoning patterns LLMs have been trained on. Semantic similarity to ground-truth facts or oracle advice from another LLM instance is used to restrict the search space and validate the traces of justification steps returned as focussed and trustable answers. At the end, the unique minimal model of a generated Horn Clause program collects the results of the reasoning process. As applications, we sketch implementations of consequence predictions, causal explanations, recommendation systems and topic-focussed exploration of scientific literature.
the knowledge representation process in Computer Music is an essential element for the development of systems. Methods have been applied to provide the computer withthe ability to infer information from previously es...
详细信息
ISBN:
(纸本)9783031085451;9783031085444
the knowledge representation process in Computer Music is an essential element for the development of systems. Methods have been applied to provide the computer withthe ability to infer information from previously established experience and definitions. Alongside this process, with regard to the aspect of musical interaction, it is observed that in the generation of automatic musical compositions, the harmonic component is fundamental in the production of associations aimed at human interaction. However, there is a lack of studies that deal withthe functions performed by chords. In this sense, Inductive logicprogramming is a growing field of research that incorporates concepts of logicprogramming and Machine Learning. this work consists in the application of the technique in Machine Learning of Inductive logicprogramming, performing the derivation of harmonic functions. Musical rules are represented by First-Order logic and a corpus-based on functional Harmony theory is used as background knowledge. through the experimental method, the induction of 3 functional rules was performed. the measures of precision, coverage, accuracy, and execution time indicate the feasibility of this approach according to the purpose of algorithmic musical composition.
While there has been much cross-fertilization between functional and logicprogramming-e.g., leading to functional models of many Prolog features-this appears to be much less the case regarding probabilistic programmi...
详细信息
ISBN:
(数字)9783030994617
ISBN:
(纸本)9783030994600;9783030994617
While there has been much cross-fertilization between functional and logicprogramming-e.g., leading to functional models of many Prolog features-this appears to be much less the case regarding probabilistic programming, even though this is an area of mutual interest. Whereas functionalprogramming often focuses on modeling probabilistic processes, logicprogramming typically focuses on modeling possible worlds. these worlds are made up of facts that each carry a probability and together give rise to a distribution semantics. the latter approach appears to be little-known in the functionalprogramming community. this paper aims to remedy this situation by presenting a functional account of the distribution semantics of probabilistic logicprogrammingthat is based on possible worlds. We present a term monad for the monadic syntax of queries together with a natural interpretation in terms of boolean algebras. then we explain that, because probabilities do not form a boolean algebra, they-and other interpretations in terms of commutative semirings-can only be computed after query normalisation to deterministic, decomposable negation normal form (d-DNNF). While computing the possible worlds readily gives such a normal form, it suffers from exponential blow-up. Using heuristic algorithms yields much better results in practice.
Various program verification problems for functional programs can be reduced to the validity checking problem for formulas of a fixpoint logic. Recently, Kobayashi et al. have shown that the unfold/fold transformation...
详细信息
ISBN:
(数字)9783030994617
ISBN:
(纸本)9783030994600;9783030994617
Various program verification problems for functional programs can be reduced to the validity checking problem for formulas of a fixpoint logic. Recently, Kobayashi et al. have shown that the unfold/fold transformation originally developed for logicprogramming can be extended and applied to prove the validity of fixpoint logic formulas. In the present paper, we refine their unfold/fold transformation, so that each predicate can be unfolded a different number of times in an asynchronous manner. Inspired by the work of Lee et al. on size change termination, we use a variant of size change graphs to find an appropriate number of unfoldings of each predicate. We have implemented an unfold/fold transformation tool based on the proposed method, and evaluated its effectiveness.
the verification framework PPV (Probabilistic Program Verification) verifies functional probabilistic programs supporting higher-order functions, continuous distributions, and conditional inference. PPV is based on th...
详细信息
ISBN:
(数字)9783030994617
ISBN:
(纸本)9783030994600;9783030994617
the verification framework PPV (Probabilistic Program Verification) verifies functional probabilistic programs supporting higher-order functions, continuous distributions, and conditional inference. PPV is based on the theory of quasi-Borel spaces which is introduced to give a semantics of higher-order probabilistic programming languages with continuous distributions. In this paper, we formalize a theory of quasi-Borel spaces and a core part of PPV in Isabelle/HOL. We first construct a probability monad on quasi-Borel spaces based on the Giry monad in the Isabelle/HOL library. Our formalization of PPV is extended so that integrability of functions can be discussed formally. Finally, we prove integrability and convergence of the Monte Carlo approximation in our mechanized PPV.
the generation of comprehensible explanations is an essential feature of modern artificial intelligence systems. In this work, we consider probabilistic logicprogramming, an extension of logicprogramming which can b...
详细信息
ISBN:
(数字)9783030994617
ISBN:
(纸本)9783030994600;9783030994617
the generation of comprehensible explanations is an essential feature of modern artificial intelligence systems. In this work, we consider probabilistic logicprogramming, an extension of logicprogramming which can be useful to model domains with relational structure and uncertainty. Essentially, a program specifies a probability distribution over possible worlds (i.e., sets of facts). the notion of explanation is typically associated withthat of a world, so that one often looks for the most probable world as well as for the worlds where the query is true. Unfortunately, such explanations exhibit no causal structure. In particular, the chain of inferences required for a specific prediction (represented by a query) is not shown. In this paper, we propose a novel approach where explanations are represented as programs that are generated from a given query by a number of unfolding-like transformations. Here, the chain of inferences that proves a given query is made explicit. Furthermore, the generated explanations are minimal (i.e., contain no irrelevant information) and can be parameterized w.r.t. a specification of visible predicates, so that the user may hide uninteresting details from explanations.
We propose a novel approach to stream definition and manipulation. Our solution is based on two key ideas. Regular corecursion, which avoids non termination by detecting cyclic calls, is enhanced, by allowing in equat...
详细信息
ISBN:
(数字)9783030994617
ISBN:
(纸本)9783030994600;9783030994617
We propose a novel approach to stream definition and manipulation. Our solution is based on two key ideas. Regular corecursion, which avoids non termination by detecting cyclic calls, is enhanced, by allowing in equations defining streams other operators besides the stream constructor. In this way, some non-regular streams are definable. Furthermore, execution includes a runtime check to ensure that the stream generated by a function call is well-defined, in the sense that access to an arbitrary index always succeeds. We extend the technique beyond the simple stream operators considered in previous work, notably by adding an interleaving combinator which has a non-trivial recursion scheme.
Capturing domain knowledge is a time-consuming procedure that usually requires the collaboration of a Subject Matter Expert (SME) and a modeling expert to encode the knowledge. this situation is further exacerbated in...
详细信息
ISBN:
(数字)9783030994617
ISBN:
(纸本)9783030994600;9783030994617
Capturing domain knowledge is a time-consuming procedure that usually requires the collaboration of a Subject Matter Expert (SME) and a modeling expert to encode the knowledge. this situation is further exacerbated in some domains and applications. the SME may find it challenging to articulate the domain knowledge as a procedure or a set of rules but may find it easier to classify instance data. In the cyber-physical domain, inferring the implemented mathematical concepts in the source code or a different form of representation, such as the Resource Description Framework (RDF), is difficult for the SME, requiring particular expertise in low-level programming or knowledge in Semantic Web technologies. To facilitate this knowledge elicitation from SMEs, we developed a system that automatically generates classification and annotation rules for control concepts in cyber-physical systems (CPS). Our proposed approach leverages the RDF representation of CPS source code and generates the rules using Inductive logicprogramming and semantic technologies. the resulting rules require a small set of labeled instance data that is provided interactively by the SME through a user interface within our system. the generated rules can be inspected, iterated and manually refined.
We show the advantages of using Swift as the programming language for behaviours on the Pepper and Nao robots as used withthe RoboCup Standard Platform League and the RoboCup@Home - Social Standard Platform. We show ...
详细信息
ISBN:
(纸本)9783030986827;9783030986810
We show the advantages of using Swift as the programming language for behaviours on the Pepper and Nao robots as used withthe RoboCup Standard Platform League and the RoboCup@Home - Social Standard Platform. We show that Swift is not only incorporating modern features of object-oriented programming and functionalprogramming, but is also now a stable systems programming language that enables both high-level development as well as fine hardware control. Deterministic memory management makes Swift suitable for real-time, embedded systems, and thus for robotic applications. Moreover, we show in this paper we can apply model-driven software-development by deploying behaviours coded as executable arrangements of logic-labelled finite-state machines (LLFSMs). We also show LLFSMs are not only suitable for reactive architectures, but also for deliberative architectures.
暂无评论