Designing and operating production systems and keeping them up to date at the speed of innovation to meet competition, consumer trends and sustainability requirements is a challenging task. The problem becomes even mo...
详细信息
Designing and operating production systems and keeping them up to date at the speed of innovation to meet competition, consumer trends and sustainability requirements is a challenging task. The problem becomes even more challenging when the production system is highly uncertain and data-poor. This complexity is increasingly recognised, as is the value that risk engineering can bring to the design, operation, maintenance and upgrading of production systems to avoid compromising their viability and ensure their continued efficiency. The manuscript is guided by this challenge and has a twofold objective. First, it analyses and explains the importance of adopting a logic-driven and simulation-based risk engineering approach to support decisionmaking. This objective is achieved by going to the root of the methodological constructs that characterise the methodologies available in the literature and by highlighting the main limitations. Second, it applies the HoRAM method to the use case of food banks to verify its suitability. The use case of food banks was chosen because it is characterised by more and unusual uncertainties compared to conventional production systems (such as uneven labour and raw material availability). The main contributions of this research are as follows. First, it provides a critical analysis of currently available risk assessment methodologies. It highlights the weaknesses and explains the implications of these weaknesses for decision support. Second, it concerns the development of a complete risk engineering process, offering an approach that goes beyond conventional risk assessment approaches that stop at the identification of the critical elements associated with the problem at hand. It explains the importance of closing the risk analysis loop by quantitatively verifying the efficiency of the identified mitigation solutions, thus moving from risk assessment to risk engineering. The results suggest that the proposed approach is suitable for suppo
Modern Machine Learning (ML) models have a significant number of hyper-parameters that need adjusting to leverage performance and energy efficiency for a given model configuration during training. This becomes a consi...
详细信息
ISBN:
(纸本)9798350344776
Modern Machine Learning (ML) models have a significant number of hyper-parameters that need adjusting to leverage performance and energy efficiency for a given model configuration during training. This becomes a considerable design challenge with increasing complexity requiring larger models. This paper explores the Tsetlin Machine (TM) - a new logic-based ML approach with only four hyper-parameters regardless of the problem space. Two of these hyper-parameters influence the TM architecture while the remaining two impact the learning efficacy. This work focuses on the systematic search for optimal hyper-parameters for the TM and aims to understand how hyper-parameter values affect performance and prediction accuracy using MNIST dataset as a case study.
Tsetlin Machine (TM) is a recent automaton-based algorithm for reinforcement learning. It has demonstrated competitive accuracy on many popular benchmarks while providing a natural interpretability. Due to its logical...
详细信息
作者：
Zerpa, LevisYachay Tech
Escuela Ciencias Sociales & Innovac San Miguel De Urcuqui Ecuador
In this paper I discuss a goal-directed framework for reasoning in first-order logicbased on conditions. It is proposed an interpretation of these conditions in terms of proof strategies that lead to the generation o...
详细信息
ISBN:
(纸本)9781509041718
In this paper I discuss a goal-directed framework for reasoning in first-order logicbased on conditions. It is proposed an interpretation of these conditions in terms of proof strategies that lead to the generation of automated advice based on a proof plan. The examined framework is compared with the well-known program Symlog.
Michal Tyburski, Circumscription: formalization of non-monotonic reasoning in second order logic We discuss circumscription, a logical formalization of non-monotonic reasoning, introduced by John McCarthy and Vladimir...
详细信息
Michal Tyburski, Circumscription: formalization of non-monotonic reasoning in second order logic We discuss circumscription, a logical formalization of non-monotonic reasoning, introduced by John McCarthy and Vladimir Lifschitz. First section contains presentation of assumptions of logic-based artificial intelligence, problem of non-monotonicity in commonsense reasoning and informal formulation of circumscription. In section two, a formal definition of circumscription is given. The idea of circumscription is discussed from syntactic and semantic point of view. Theoretical investigations are supplemented with examples. In section three, methods of computing circumscription are discussed. Section four contains exemplary circumscription-based formalization of simple non-monotonic reasoning. Finally, not only a comment about the role of logic in artificialintelligence is made but a piece of information about implementation of circumscription is given as well.
暂无评论