Call for papers
The ECML-PKDD 2017 Workshop on Automatic Machine Learning (AutoML)
Collocated with ECML-PKDD in Skopje, Macedonia, September 22, 2017
Submission deadline: 16 July, 2017 (passed)
Notification: 30 July, 2017
Paper camera-ready: Monday, 21 August, 2017 (new date)
Late breaking / PhD student papers:
Submission deadline: 25 August 2017
Notification: 1 September, 2017
AutoML: Automatic selection, configuration and composition of machine learning algorithms
This workshop will provide a platform for discussing recent developments in the areas of meta-learning, algorithm selection and configuration, which arise in many diverse domains and are increasingly relevant today. Researchers and practitioners from all areas of science and technology face a large choice of parameterized machine learning algorithms, with little guidance as to which techniques to use in a given application context. Moreover, data mining challenges frequently remind us that algorithm selection and configuration are crucial in order to achieve cutting-edge performance, and drive industrial applications. Meta-learning leverages knowledge of past algorithm applications to select the best techniques for future applications, and offers effective techniques that are superior to humans both in terms of the end result and especially in the time required to achieve it. In this workshop, we will discuss different ways of exploiting meta-learning techniques to identify the potentially best algorithm(s) for a new task, based on meta-level information, including prior experiments on both past datasets and the current one. Many contemporary problems also require the use of complex workflows that consist of several processes or operations. Constructing such complex workflows requires extensive expertise, and could be greatly facilitated by leveraging planning, meta-learning and intelligent system design. This task is inherently interdisciplinary, as it builds on expertise in various areas of AI.
Main research areas of relevance to this workshop include, but are not limited to:
- Algorithm / model selection and configuration
- Meta-learning and exploitation of meta-knowledge
- Hyperparameter optimization
- Automatic generation and evaluation of learning processes / workflows
- Representation learning and automatic feature extraction / construction
- Automatic feature coding / transformation
- Automatic detection and handling skewed data or missing values
- Automatic acquisition of new data (active learning, experimental design)
- Usage of planners in the construction of workflows
- Reinforcement learning for parameter control & algorithm design
- Representation of learning goals and states in learning
- Control and coordination of learning processes
- Layered learning
- Multi-task and transfer learning
- Learning to learn
- Intelligent experiment design
Invited speaker: Michele Sebag, CNRS, France
Co-chairs: Frank Hutter, Holger Hoos, Pavel Brazdil and Joaquin Vanschoren
There are two options:
The standard option: up to 6 pages (not including references) in ECML-PKDD format.
The non-standard option: longer papers of up to 15 pages (not including references) in ECML-PKDD format.
In case of doubt, please use the standard option of 6 pages. Out of two papers with similar content spread over 6 or 15 pages, we would clearly prefer the shorter version. The non-standard option is only available to minimize overhead for authors that have on-topic 15-page papers ready to go.
For details on how to submit, please see the submission page; the submission deadline is July 10th, 2017.
All accepted papers will be presented as posters and very short poster spotlights; the best paper(s) will be selected for an oral presentation.
At least one author of each accepted paper should be registered for the main conference.
The organizers are planning to publish all accepted submissions as CEUR proceedings and the authors of accepted papers will get more details about the deadlines etc. later.
Late breaking and/or PhD student papers
These submissions should be limited to 2 pages, including references and should be submitted via EasyChair.
They will undergo the usual, but rather quick reviewing process.