Re: [UAI] A perplexing problem - Version 2
On Mon, 23 Feb 2009, Francisco Javier Diez wrote:
> Konrad Scheffler wrote:
> > I agree this is problematic - the notion of calibration (i.e. that you can
> > say P(S|"70%") = .7) does not really make sense in the subjective Bayesian
> > framework where different individuals are working with different priors,
> > because different individuals will have different posteriors and they can't
> > all be equal to 0.7.
>
> I apologize if I have missed your point, but I think it does make sense. If
> different people have different posteriors, it means that some people will
> agree that the TWC reports are calibrated, while others will disagree.
I think this is another way of saying the same thing - if you define the
concept of calibration such that people will, depending on their priors,
disagree over whether the reports are calibrated then it is still
problematic to prescribe calibration in the problem formulation - because
this will mean different things to different people. Unless you take
"TWC is calibrated" to mean "everyone has the same prior as TWC", which I
don't think was the intention in the original question.
In my opinion the source of confusion here is the use of a subjective
Bayesian framework (i.e. one where the prior is not explicitly stated and
is assumed to be different for different people). If instead we use an
objective Bayesian framework where all priors are stated explicitly, the
difficulties disappear.
> Who is right? In the case of unrepeatable events, this question would not make
> sense, because it is not possible to determine the "true" probability, and
> therefore whether a person or a model is calibrated or not is a subjective
> opinion (of an external observer).
>
> However, in the case of repeatable events--and I acknowledge that
> repeatability is a fuzzy concept--, it does make sense to speak of an
> objective probability, which can be identified with the relative frequency.
> Subjective probabilities that agree with the objective probability (frequency)
> can be said to be correct and models that give the correct probability for
> each scenario will be considered to be calibrated.
>
> If we accept that "snow" is a repeatable event, the all the individuals should
> agree on the same probability. If it is not, P(S|"70%") may be different for
> each individual because having different priors and perhaps different
> likelihoods or even different structures in their models.
I strongly disagree with this. The ("true") relative frequency is not the
same thing as the correct posterior. One can imagine a situation where the
correct posterior (calculated from the available information) is very far
from the relative frequency which one would obtain given the opportunity
to perform exhaustive experiments.
Probabilities (in any variant of the Bayesian framework) do not describe
reality directly, they describe what we know about reality (typically in
the absence of complete information).
> Coming back to the main problem, I agree again with Peter Szolovits in making
> the distinction between likelihood and posterior probability.
>
> a) If I take the TWC forecast as the posterior probability returned by a
> calibrated model (the TWC's model), then I accept that the probability of snow
> is 70%.
>
> b) However, if I take "70% probability of snow" as a finding to be introduced
> in my model, then I should combine my prior with the likelihood ratio
> associated with this finding, and after some computation I will arrive at
> P(S|"70%") = 0.70. [Otherwise, I would be incoherent with my assumption that
> the model used by the TWC is calibrated.]
>
> Of course, if I think that the TWC's model is calibrated, I do not need to
> build a model of TWC's reports that will return as an output the same
> probability estimate that I introduce as an input.
>
> Therefore I see no contradiction in the Bayesian framework.
But this argument only considers the case where your prior is identical
to TWC's prior. If your prior were _different_ from theirs (the more
interesting case) then you would not agree that they are calibrated.
___
uai mailing list
uai@ENGR.ORST.EDU
https://secure.engr.oregonstate.edu/mailman/listinfo/uai
[UAI] Second CFP: The IJCAI-09 Workshop on Learning Structural Knowledge from Observations (StrucK-09)
*** Our apologies if you received this announcement multiple times CALL FOR PAPERS: StrucK-09: The IJCAI-09 Workshop on Learning Structural Knowledge from Observations Pasadena, California, USA, July 12, 2009, http://www.cs.umd.edu/users/ukuter/struck09/ to be held in conjunction with the 21st International Joint Conference on Artificial Intelligence (IJCAI-09) * Paper Submission Deadline: March 6, 2009 * ** DESCRIPTION: Human cognition organizes knowledge in different complexity levels: higher-level knowledge is formed by first acquiring simple concepts, which are then combined to learn complex ones. As a result, many cognitive architectures use structural models to represent relations between knowledge of different complexity. Structural modeling has led to a number of representation and reasoning formalisms including frames, schemas, abstractions, hierarchical task networks (HTNs), and goal graphs among others. These formalisms have in common the use of certain kinds of constructs (e.g., objects, goals, skills, and tasks) that represent knowledge of varying degrees of complexity and that are connected through structural relations. In recent years, we have observed increasing interest towards the problem of learning such structural knowledge from observations. These observations range from traces generated by an automated planner to video feeds from a robot performing some actions. Researchers have been addressing instances of this problem from different perspectives in a variety of research communities, among others including -- Machine Learning (including inductive logic programming (ILP)) -- Automated Planning -- Case-Based Reasoning -- Cognitive Science We believe that the time is ripe to get together researchers from these and other communities that are looking into instances of this problem and share ideas and perspectives in a common forum. Potential focus topics include but are not limited to: - Cognitive architectures and learning techniques such as ILP, explanation-based learning (EBL), abstraction, generalization, and teleoreactive logic programs - Formalisms for goal-directed behavior, including hierarchical task networks, skill hierarchies, goal networks, and annotated goal hierarchies - Learning behavior from observations over time - Observations ranging from fully to partially observable inputs and from annotated to un-annotated action traces - Trade-offs between task performance and structural learning - Learning meta-level knowledge (i.e., how to choose among different reasoning/problem-solving functionalities, how to manage the trade-offs between task performance and learning) - Probabilistic and other extensions to structural knowledge to represent uncertainty - Representing and learning continuous information - Interacting with the external environment during structural learning (i.e., information-gathering, execution, etc) - Learning structural information/data flow from observations ** IMPORTANT DATES: Paper Submission: March 6, 2009 * Notifications of Acceptance/Rejection: April 17, 2009 Camera-Ready Papers: May 8, 2009 Announcement of the Workshop Program: May 22, 2009 Workshop Date: July 12, 2009 ** SUBMISSION INSTRUCTIONS: Paper submissions must be formatted in IJCAI style (see: http://ijcai-09.org/fcfp.html for instructions and to download file templates). We solicit paper ranging from 2 pages (extended abstracts) to 6 pages (full papers). The submissions will be on the EasyChair Conference Management System (http://www.easychair.org/ ). You may access the workshop site at EasyChair by first logging onto EasyChair at the following URL: https://www.easychair.org/login.cgi You must click on the "StrucK-09" link that will appear when you logged in. If you do not have an EasyChair username and password, then you may get one at the following URL: http://www.easychair.org/conferences/account_apply.cgi If you experience any problems with logging in and using EasyChair paper submissions, please let us know at struc...@easychair.org. ** WORKSHOP PROGRAM: The workshop will interleave short presentations, a poster session, two or more discussion groups, and a joint open discussion session. All accepted papers will be asked to present a poster; selected participants will be invited to give 15 minute overviews unless they express a preference not to. The workshop is aimed to identifying and discussing specific questions that are still open and/or that are still prone to further understanding and research in order to develop efficient and intelligent systems. To achieve this objective, we plan organize break-out working groups during the course of the workshop. Each break-out group will focus on one specific topic in Learning Structural Knowledge
[UAI] Call for Participation: EACL 2009 workshop on Computational Linguistic Aspects of Grammatical Inference
Apologies to those of you who receive this more than once. EACL 2009 workshop on Computational Linguistic Aspects of Grammatical Inference 2nd Call for Participation 27 February 2009: early bird registration deadline 30 March 2009 Co-located with The 12th Conference of the European Chapter of the Association for Computational Linguistics Athens, Greece http://ilk.uvt.nl/clagi09 Scope There has been growing interest over the last few years in learning grammars from natural language text (and structured or semi-structured text). The family of techniques enabling such learning is usually called "grammatical inference" or "grammar induction". The field of grammatical inference is often subdivided into formal grammatical inference, where researchers aim to proof efficient learnability of classes of grammars, and empirical grammatical inference, where the aim is to learn structure from data. In this case the existence of an underlying grammar is just regarded as a hypothesis and what is sought is to better describe the language through some automatically learned rules. Both formal and empirical grammatical inference have been linked with (computational) linguistics. Formal learnability of grammars has been used in discussions on how people learn language. Some people mention proofs of (non-)learnability of certain classes of grammars as arguments in the empiricist/nativist discussion. On the more practical side, empirical systems that learn grammars have been applied to natural language. Instead of proving whether classes of grammars can be learnt, the aim here is to provide practical learning systems that automatically introduce structure in language. Example fields where initial research has been done are syntactic parsing, morphological analysis of words, and bilingual modeling (or machine translation). This workshop at EACL 2009 aims to explore the state-of-the-art in these topics. In particular, we aim at bringing formal and empirical grammatical inference researchers closer together with researchers in the field of computational linguistics. Preliminary Programme Session 1 09:00-09:30 Introduction 09:30-10:30 Damir Äavar Invited talk: On bootstrapping of linguistic features for bootstrapping grammars Session 2: Transduction 11:00-11:30 Jeroen Geertzen Dialogue Act Prediction Using Stochastic Context-Free Grammar Induction 11:30-12:00 Dana Angluin and Leonor Becerra-Bonache Experiments Using OSTIA for a Language Production Task 12:00-12:30 Jorge González and Francisco Casacuberta GREAT: a finite-state machine translation toolkit implementing a Grammatical Inference Approach for Transducer Inference (GIATI) Session 3: Language models and parsing 14:00-14:30 Alexander Clark, Remi Eyraud and Amaury Habrard A note on contextual binary feature grammars 14:30-15:00 Herman Stehouwer and Menno van Zaanen Language models for contextual error detection and correction 15:00-15:30 Marie-HélÚne Candito, Benoit Crabbé and Djamé Seddah On statistical parsing of French with supervised and semi-supervised strategies 15:30-16:00 Franco M. Luque and Gabriel Infante-Lopez Upper Bounds for Unsupervised Parsing with Unambiguous Non-Terminally Separated Grammars Session 4: Morphology 16:30-17:00 Katya Pertsova A comparison of several learners for Boolean partitions: implications for morphological paradigm 17:00-18:00 Panel discussion Registration Registration for the workshop is done through the EACL website. http://www.eacl2009.gr/conference/callforparticipation points to the EACL Call for Participation. Registration can be done at http://www.eacl2009.gr/conference/registration Programme Committee Pieter Adriaans, University of Amsterdam, The Netherlands Srinivas Bangalore, AT&T Labs-Research, USA Leonor Becerra-Bonache, Yale University, USA Rens Bod, University of Amsterdam, The Netherlands Antal van den Bosch, Tilburg University, The Netherlands Alexander Clark, Royal Holloway, University of London, UK Walter Daelemans, University of Antwerp, Belgium Shimon Edelman, Cornell University, USA Jeroen Geertzen, University of Cambridge, UK Jeffrey Heinz, University of Delaware, USA Colin de la Higuera, Université de Saint-Etienne, France (co-chair) Alfons Juan, Universidad Politecnica de Valencia, Spain Frantisek Mraz, Charles University, Czech Republic Khalil Sima'an, University of Amsterdam, The Netherlands Richard Sproat, University of Illinois at Urbana-Champaign, USA Menno van Zaanen, Tilburg University, The Netherlands (co-chair) Willem Zuidema, University of Amsterdam, The Netherlands Organizing Committee Menno van Zaanen, Tilburg University, The Netherlands (co-chair) Colin de la Higuera, Université de Saint-Etienne, France (co
[UAI] Internships at Siemens Healthcare
Internship in Natural language Processing, Machine Learning / Data Mining Computer-Aided Diagnosis (CAD) & Knowledge Solutions Group (CKS) Siemens Healthcare, Malvern, PA, USA The CKS Group at Siemens Healthcare, USA, is building up a world-class R&D team that includes software engineers and research scientists in machine learning, computer vision, probabilistic inference, and natural language processing. We are looking for outstanding interns for research and software development in natural language processing and machine learning with an emphasis on information extraction from unstructured text. As an intern, you will join our team of scientists in solving exciting and challenging research problems in the medical and biomedical fields. Our research is motivated by decision-support and data processing problems arising in the medical domain and related health areas; experience or interest in these areas is a plus. Our team currently conducts research in Bayesian methods, probabilistic inference, statistical learning theory, optimization, statistics, natural language processing, data mining and works closely with a team of image processing scientists. Qualifications: a) Pursuing a Bachelor/Master/PhD in computer science/statistics/engineering or related discipline. b) Software skills must include languages like MATLAB/C++/C#/Java, scripting languages (Perl/Python), and working with databases (SQL). c) Experience developing web applications (knowledge of AJAX, Java Servlets, and GWT) is a plus d) Excellent communication skills. e) Knowledge of data mining / machine learning / image processing/ medical domain is a plus. You will be expected to spend at least 10-12 weeks; however there is much flexibility in the starting and finishing dates (non-summer and longer internships are also considered). We also consider international interns. In order to apply, please follow these steps: 1) Email your CV and contact number to the address below. Highlight your coding/research skills and mention the largest size projects you have worked on. 2) Request one letter of recommendation (preferably from your advisor) to be emailed to us. 3) Briefly (e.g., in half a page) tell us about you research interests. Highlight any prior machine learning/modeling experience with real-life datasets. Email the above to vikas.ray...@siemens.com, with the subject "Siemens Internships". Siemens HealthCare is located in Malvern, PA. Malvern lies within the Main Line area of the Philadelphia suburbs (with downtown Philly less than an hour's drive away). This message and any included attachments are from Siemens Medical Solutions and are intended only for the addressee(s). The information contained herein may include trade secrets or privileged or otherwise confidential information. Unauthorized review, forwarding, printing, copying, distributing, or using such information is strictly prohibited and may be unlawful. If you received this message in error, or have reason to believe you are not authorized to receive it, please promptly delete this message and notify the sender by e-mail with a copy to central.securityoff...@siemens.com Thank you ___ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
[UAI] Fusion'09 Paper Submission Deadline Extended
Dear Colleagues, Please accept our apology if you receive multiple copies of this announcement. The deadline for paper submission to Fusion'09 has been extended. 12th International Conference on Information Fusion Grand Hyatt, Seattle Washington, USA, 6-9 July 2009 http://www.fusion2009.org Important Deadlines: Papers submission 2 March 2009 16 March 2009 Acceptance of papers10 April 2009 17 April 2009 Final papers15 May 2009 Early registration 15 May 2009 Overview - The 12th International Conference on Information Fusion will be held in Seattle, Washington, USA, at the Grand Hyatt Seattle Hotel. Authors are invited to submit papers describing advances and applications in information fusion, with submission of non-traditional topics encouraged. Conference Site - Pacific Northwest is one of the most scenic parts of United States and Seattle is the home of some of the world's biggest technology companies such as Boeing and Microsoft. Seattle is easily accessible from other parts of North America, Europe, and the Asia Pacific region. The Grand Hyatt Seattle has been selected as the site for Fusion 2009. Nestled conveniently in the heart of downtown's thriving retail and theatre district, Grand Hyatt Seattle redefines luxury, design and technology. Organizing Committee: General Co-Chairs Robert Lynch, Naval Undersea Warfare Center, USA Chee-Yee Chong, BAE Systems, USA Executive Chair Robert Lobbia, Boeing, USA Technical Co-Chairs Kuo-Chu Chang, George Mason University, USA Mahendra Mallick, Georgia Institute of Technology, USA Finance Chair Dale Blair, Georgia Institute of Technology, USA Administrative Chair Tammy Blair, Georgia Institute of Technology, USA Publications Chair Wayne Blanding, York College of Pennsylvania, USA Tutorials/Awards Chair Ivan Kadar, Interlink Systems Sciences, Inc., USA Special Sessions Chair Chris Hempel, Naval Undersea Warfare Center, USA Publicity Chair Shanchieh Jay Yang, Rochester Institute of Technology, USA Contact Information Robert Lynch lync...@npt.nuwc.navy.mil 401-832-8663 Chee Chong chee.ch...@baesystems.com 650-210-8822 Topics of interest include (but not limited to) the following: 1. Foundational tools Probability theory; Bayesian, neural, set-membership and Dempster-Shafer approaches; random sets; fuzzy logic; risk-sensitive approaches; fusion modeling; agents; genetic optimization. 2. Technological advances Sensor modeling (radar, active and passive sonar, acoustic, seismic, magnetic, optical, visual); fusion-related hardware, software and communications technology. 3. Algorithmic developments Classification; data mining; nonlinear filtering and smoothing; target tracking and localization; contact-based tracking algorithms; combined detection/tracking; sensor resource management; distributed fusion; active and passive data fusion; data registration; image fusion; database fusion. 4. Application Areas Situation awareness and decision support in defense and intelligence; public security; aerospace; computer security; robotics and automation; intelligent transportation; logistics; automotive; manufacturing; economics and financial; environmental monitoring; medical care, etc.. Paper Submissions. Prospective authors are invited to submit papers electronically through the conference website (http://www.fusion2009.org), where paper templates and submission instructions will be available in late 2008. Papers submissions are due by 2 March 2009 and cannot exceed 8 pages. Student Paper Program. Fusion 2009 is featuring a student paper program to encourage the involvement of young engineers and scientists in information fusion. Conference fees will be discounted for the lead (student) authors of the papers. Further details are available at the conference website. ___ Shanchieh Jay Yang Associate Professor Department of Computer Engineering Rochester Institute of Technology Room 09-3425, (O) 1-585-475-6434 http://www.ce.rit.edu/people/yang ___ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
Re: [UAI] A perplexing problem - Last Version
Dear Paul, Bayesian inference is still appropriate for both problems. There are two issues here: 1) the subjectivist Bayesian viewpoint is confusing because it does not make it explicit on which information you are conditioning when setting up your prior - it becomes much clearer if you use an objective Bayesian framework (see below). 2) You are describing a situation where your two sources of information are dependent, but you are not quantifying the dependency. As Jean-Louis points out, the problem becomes simple if you are prepared to make an independence assumption (but I think this avoids the difficulty you are asking about: "In part they are using the same background knowledge that Analyst A has"). Below I give the full solution (which unfortunately is only useful if you can quantify the dependencies - for this you need a model of how the analysts go about calculating their reported probabilities). I'll use "X" as a shorthand for the statement "X is at location Y". Let's assume all analysts give their statements as a numerical value which quantifies their confidence that X is true. Let's call the value provided by the spectral analyst B and that provided by the chemical analyst C. Let I denote the information available to analyst A before reading the reports, so that her prior for X is P(X|I). We want to know P(X|BCI), which can be written as follows: P(X|BCI) = P(B|XCI)P(X|CI)/P(B|CI) where P(X|CI) = P(C|XI)P(X|I)/[P(C|XI)P(X|I)+P(C|(not X)I)P((not X)|I)] and P(B|CI) = P(B|XCI)P(X|CI) + P(B|(not X)CI)P((not X)|CI). The quantities P(X|I), P((not X)|I), P(C|XI) and P(C|(not X)I) are straightforward, but instead of P(B|XI) and P(B|(not X)I) we need to know P(B|XCI) and P(B|(not X)CI). Once these are known the answer follows. Hope this is useful, Konrad On Thu, 19 Feb 2009, Lehner, Paul E. wrote: > Austin, Jean-Lous, Konrad, Peter > > Thank you for your responses. They are very helpful. > > Your consensus view seems to be that when receiving evidence in the form > of a single calibrated judgment, one should not update personal > judgments by using Bayes rule. This seems incoherent (from a strict > Bayesian perspective) unless perhaps one explicitly represents the > overlap of knowledge with the source of the calibrated judgment (which > may not be practical.) > > Unfortunately this is the conclusion I was afraid we would reach, > because it leads me to be concerned that I have been giving some bad > advice about applying Bayesian reasoning to some very practical > problems. > > Here is a simple example. > > Analyst A is trying to determine whether X is at location Y. She has > two principal evidence items. The first is a report from a spectral > analyst that concludes "based on the match to the expected spectral > signature I conclude with high confidence that X is at location Y". > The second evidence is a report from a chemical analyst who asserts, > "based on the expected chemical composition that is typically associated > with X, I conclude with moderate confidence that X is at location Y." > How should analyst A approach her analysis? > > Previously I would have suggested something like this. "Consider each > evidence item in turn. Assume that X is at location Y. What are the > chances that you would receive a 'high confidence' report from the > spectral analyst, ... a report of 'moderate confidence' from the > chemical analyst. Now assume X is not a location Y, " In other > words I would have lead the analyst toward some simple instantiation of > Bayes inference. > > But clearly the spectral and chemical analyst are using more than just > the sensor data to make their confidence assessments. In part they are > using the same background knowledge that Analyst A has. Furthermore > both the spectral and chemical analysts are good at their job, their > confidence judgments are reasonably calibrated. This is just like the > TWC problem only more complex. > > So if Bayesian inference is inappropriate for the TWC problem, is it also > inappropriate here? Is my advice bad? > > Paul > > > From: uai-boun...@engr.orst.edu [mailto:uai-boun...@engr.orst.edu] On Behalf > Of Lehner, Paul E. > Sent: Monday, February 16, 2009 11:40 AM > To: uai@ENGR.ORST.EDU > Subject: Re: [UAI] A perplexing problem - Version 2 > > UAI members > > Thank you for your many responses. You've provided at least 5 distinct > answers which I summarize below. > (Answer 5 below is clearly correct, but leads me to a new quandary.) > > > > Answer 1: "70% chance of snow" is just a label and conceptually should be > treated as "XYZ". In other words don't be fooled by the semantics inside the > quotes. > > > > My response: Technically correct, but intuitively unappealing. Although I > often council people on how often intuition is misleading, I just couldn't > ignore my intuition on this one. > > > >
[UAI] ESANN'2009 programme
(Our apologies if you get multiple copies of this message, despite our precautions) = ESANN'2009 17th European Symposium on Artificial Neural Networks Advances in Computational Intelligence and Learning Bruges (Belgium) - April 22-23-24, 2009 Preliminary program = The preliminary program of the ESANN'2009 conference is now available on the Web: http://www.dice.ucl.ac.be/esann For those of you who maintain WWW pages including lists of related machine learning and artificial neural networks sites: we would appreciate if you could add the above URL to your list; thank you very much! For 17 years the ESANN conference has become a major event in the field of neural computation and machine learning. ESANN is a selective conference focusing on fundamental aspects of artificial neural networks, machine learning, statistical information processing and computational intelligence. Mathematical foundations, algorithms and tools, and applications are covered. This year, around 100 scientific communications will be presented, covering most areas of the neural computation and related fields. The program of the conference can be found at http://www.dice.ucl.ac.be/esann, together with practical information about the conference venue, registration, etc. Other information can be obtained by sending an e-mail to es...@uclouvain.be. ESANN - European Symposium on Artificial Neural Networks - Advances in Computational Intelligence and Learning http://www.dice.ucl.ac.be/esann * For submissions of papers, reviews, registrations: Michel Verleysen Univ. Cath. de Louvain - Machine Learning Group 3, pl. du Levant - B-1348 Louvain-la-Neuve - Belgium tel: +32 10 47 25 51 - fax: + 32 10 47 25 98 mailto:es...@uclouvain.be * Conference secretariat d-side conference services 24 av. L. Mommaerts - B-1140 Evere - Belgium tel: + 32 2 730 06 11 - fax: + 32 2 730 06 00 mailto:es...@uclouvain.be ___ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
[UAI] 7th International Planning Competition: Call for Bids
[Apologies for cross-posting] The International Planning Competition Committee is seeking bids from potential organisers of the 7th IPC, to complete alongside ICAPS 2010. Deadline: Proposals to be sent to Derek Long (de...@cis.strath.ac.uk) by the end of March, 2009. The IPC series has been highly influential in promoting the development and evaluation of planning systems, including pushing the broad research community to examine increasingly ambitious planning problems as the series has progressed. The IPC series has been significant in achieving a situation in which there are several easily available planners capable of solving large and sophisticated planning problems within reasonable performance parameters. The goal of the IPC series is to promote development and evaluation of planning technology by: a) Providing a growing repository of benchmark problems for evaluating and testing planning systems b) Developing new domains and problems that are increasingly realistic and relevant and push the state of the art in planning technology c) Encouraging direct comparison of planning systems and techniques through regular competitions d) Continuing to develop and improve a common language for expressing planning domains and problems e) Encouraging researchers and competitors to make planning software and tools publicly available We wish to encourage broad participation in these activities by both the established community and by relative newcomers like graduate students. For this reason, our objective is to keep the focus of competitions on new challenges, so that there is not a significant barrier for new participants, and so that the work done does not degenerate into purely software engineering, optimization, and tuning. Previous organisers include: 1st IPC: Drew McDermott 2nd IPC: Fahiem Bacchus 3rd IPC: Maria Fox and Derek Long 4th IPC: Deterministic: Stefan Edelkamp and Joerg Hoffmann Probabilistic: Michael Littman and Hakan Younes 5th IPC: Alfonso Gerevini with Yannis Dimopoulos, Patrik Haslum and Alessandro Saetti Probabilistic: Blai Bonet and Bob Givan 6th IPC: Deterministic: Malte Helmert, Minh Binh Do and Yannis Dimopolous Probabilistic: Dan Bryce and Olivier Buffet Learning: Alan Fern, Roni Khardon and Prasad Tadepalli Proposers should not feel constrained to follow the tracks that have been used in the past. The responsibilities of the organisers are to establish the objectives of the competition, assemble appropriate domains, provide the environment for testing and then to perform a sufficient analysis of the results to provide a brief summary at ICAPS, including identifying competitors who should be rewarded for their success. In previous competitions some organisers have sought to extend PDDL in order to express new features of planning problems. Proposals for extensions are considered very carefully and discussed and debated amongst members of the research community. The International Planning Competition Committee has been formed partly to formalise this process, where it is required. Bids should include the following information: - The name or names of the IPC7 Chair or Co-Chairs. - What tracks the proposal covers. - What are the objectives intended for IPC7 (specific track) should the bid be successful. - Whether language extensions are envisaged. - A broad indication of the expected structure of the IPC7 track: timetable, number of domains and problems, anticipated schedule for testing and the process of final evaluation. Fernando Fernández Publicity Chair ___ uai mailing list uai@ENGR.ORST.EDU https://secure.engr.oregonstate.edu/mailman/listinfo/uai
