Fig 7 - uploaded by Martin Wirsing
Content may be subject to copyright.
Download
View publication
Sample trajectories of the particle (light to dark blue, color gradient denotes time) and the agent (light to dark red).

Sample trajectories of the particle (light to dark blue, color gradient denotes time) and the agent (light to dark red).

Source publication
Fig. 1: Ensemble Development Life Cycle EDLC.
Fig. 2: AIDL life cycle.
Fig. 3: Class diagram of search-andrescue domain.
Fig. 4: Autonomous agent performance. Reward is given for victims at...
+4 Fig. 5: Autonomous agent performance despite unexpected events at...
Towards Systematically Engineering Autonomous Systems using Reinforcement Learning and Planning Draft. Final version to appear in Proc. Analysis, Verification and Transformation for Declarative Programming and Intelligent Systems (AVERTIS)
Preprint
Full-text available
  • Jul 2022
Autonomous systems need to be able dynamically adapt to changing requirements and environmental conditions without redeploy-ment and without interruption of the systems functionality. The EU project ASCENS has developed a comprehensive suite of foundational theories and methods for building autonomic systems. In this paper we specialise the EDLC pr...
Cite
Download full-text

Context in source publication

Context 1
... example, we can observe that the SNES agent learns to follow the particle closely. Figure 7 illustrates this by sample trajectories of the particle and the agent (color gradients denote time). Figure 8 shows the proportion of episodes that satisfy the given requirement. ...

Citations

... Yet, interaction points between traditional SD and self-adaptation are only loosely dened. Recently, AIDL [35] specialized the EDLC to the construction of autonomous policies using Planning and Reinforcement Learning techniques. Overall, the ASCENS approach emphasizes formal verication. ...
Capturing Dependencies Within Machine Learning via a Formal Process Model
Chapter
  • Oct 2022
The development of Machine Learning (ML) models is more than just a special case of software development (SD): ML models acquire properties and fulfill requirements even without direct human interaction in a seemingly uncontrollable manner. Nonetheless, the underlying processes can be described in a formal way. We define a comprehensive SD process model for ML that encompasses most tasks and artifacts described in the literature in a consistent way. In addition to the production of the necessary artifacts, we also focus on generating and validating fitting descriptions in the form of specifications. We stress the importance of further evolving the ML model throughout its life-cycle even after initial training and testing. Thus, we provide various interaction points with standard SD processes in which ML often is an encapsulated task. Further, our SD process model allows to formulate ML as a (meta-) optimization problem. If automated rigorously, it can be used to realize self-adaptive autonomous systems. Finally, our SD process model features a description of time that allows to reason about the progress within ML development processes. This might lead to further applications of formal methods within the field of ML.KeywordsMachine learningProcess modelSelf-adaptationSoftware engineering
View