THE 6TH SYMPOSIUM ON CONFORMAL AND PROBABILISTIC PREDICTION WITH APPLICATIONS (COPA 2017)
14-16 June, 2017
Co-organised by: Royal Holloway, University of London (UK) and Karolinska Institutet, Sweden
Quantifying the uncertainty of the predictions produced by classification and regression techniques is an important problem in the field of Machine Learning. Conformal Prediction is a framework for complementing the predictions of Machine Learning algorithms with reliable measures of confidence. The methods developed based on this framework produce well-calibrated confidence measures for individual examples without assuming anything more than that the data are generated independently by the same probability distribution (i.i.d.).
Since its development the framework has been combined with many popular techniques, such as Support Vector Machines, k-Nearest Neighbours, Neural Networks, Ridge Regression etc., and has been successfully applied to many challenging real world problems, such as the early detection of ovarian cancer, the classification of leukaemia subtypes, the diagnosis of acute abdominal pain, the assessment of stroke risk, the recognition of hypoxia in electroencephalograms (EEGs), the prediction of plant promoters, the prediction of network traffic demand, the estimation of effort for software projects and the backcalculation of non-linear pavement layer moduli. The framework has also been extended to additional problem settings such as semi-supervised learning, anomaly detection, feature selection, outlier detection, change detection in streams and active learning. The aim of this symposium is to serve as a forum for the presentation of new and ongoing work and the exchange of ideas between researchers on any aspect of Conformal Prediction and its applications.
Invited TalksProf. Vladimir Vapnik, AI Research Facebook, Columbia University USA and Royal Holloway, University of London, UK
- Intelligent Methods of Learning
The symposium welcomes submissions introducing further developments and extensions of the Conformal Prediction framework and describing its application to interesting problems of any field.
The topics of the symposium include, but are not limited to:
- Non-conformity measures
- Venn prediction
- On-line compression modeling
- Theoretical analysis of Conformal Prediction techniques
- Applications/usages of Conformal Prediction
- Machine learning
- Pattern recognition
- Regression estimation
- Density estimation
- Algorithmic information theory
- Measures of confidence
- Applications in Bioinformatics and Medicine
- Applications in Information Security and Homeland Security
- Data mining and visualization
- Big data applications
- Data analysis applications in science and engineering
- Uncertainty quantification
Special Session on Novel Directions of Applying Machine Learning in Cheminformatics
Invited SpeakerDr. Andreas Bender, Centre for Molecular Informatics, Department of Chemistry, Cambridge University, UK
- Conformal and Probabilistic Prediction for Chemical and Biological Data
Abstract More and more chemical and biological information is becoming available, both in public databases as well as in company repositories. However, how to make use of this information in chemical biology and drug discovery settings is much less clear. Here, in particular, the noise of biological measurements, as well as the different types of readouts one can choose to measure to describe a biological response, are often difficult to handle in practice. In this presentation, we will discuss how probabilistic modelling and conformal prediction can be used to model noisy biological data in different settings. Firstly, we will show how Bayesian models can be used efficiently to predicting protein targets of ligands, based on more than 195 million bioactivity data points and covering 1,080 target proteins. Secondly, we will apply conformal prediction to predict the cytotoxicity of compounds measured in different assay systems and cell lines, where it is well established that responses are not consistent between biological systems. Finally, we will apply docking in combination with conformal prediction to iterative compound screening, which enables us to select subsets of compounds for screening with a given confidence level of showing activity, and hence a tailoring to the particular resources available in a given experimental setting.
There has been a renewed interest in novel machine learning techniques in drug discovery during the last years. This has been driven both by novel methods, access to larger and imbalanced datasets as well progress in high-performance computing. For example, methods like conformal prediction, deep learning and matrix factorization have already made significant impact and are part of making the drug discovery process more data driven and efficient. In this session we will mainly focus on the cheminformatics aspects. The speakers will describe the current state of the art, bottlenecks and future directions covering topics like de novo design of novel molecules, improve accuracy in activity prediction, and confidence estimation. The presentations will be followed by round table discussions focusing on current challenges and future opportunities.
Please see more information about the special session by downloading this pdf.
Authors are invited to submit original, English-language research contributions or experience reports. Papers should be no longer than 20 pages formatted according to the well-known JMLR (Journal of Machine Learning Research) style. The LaTeX package for the style is available here. All aspects of the submission and notification process will be handled online via the EasyChair Conference System.
Authors are invited to submit abstracts for the poster sessions. The poster abstract submission deadline is 10th May, 2017 and the submission will be handled online via the EasyChair Conference System. Only registered delegates can present the poster at the conference. Unfortunately, we can only include full papers in the conference proceedings. However, we plan to place all the poster abstracts on the conference website.
Submitted papers will be refereed for quality, correctness, originality, and relevance. Notification and reviews will be communicated via email. All accepted papers will be presented at the conference and published by JMLR Workshop and Conference Proceedings (volume 60).
Accepted papers in the special session “Novel Directions of Applying Machine Learning in Cheminformatics” will be invited to submit an extended version of their manuscripts to a special issue in Journal of Cheminformatics.
- Paper Submission Deadline:
March 31st, 2017April 7th, 2017
- Poster Abstract Submission Deadline: May 5th, 2017
- Author Notifications: May 5th, 2017
- Camera-ready Submission Deadline: May 23rd, 2017
- Symposium Dates: June 14th-16th, 2017