BACKGROUND: Proteins interact through specific binding interfaces that contain many residues in domains. Protein interactions thus occur on three different levels of a concept hierarchy: whole-proteins, domains, and residues. Each level offers a distinct and complementary set of features for computationally predicting interactions, including functional genomic features of whole proteins, evolutionary features of domain families and physical-chemical features of individual residues. The predictions at each level could benefit from using the features at all three levels. However, it is not trivial as the features are provided at different granularity. RESULTS: To link up the predictions at the three levels, we propose a multi-level machine-learning framework that allows for explicit information flow between the levels. We demonstrate, using representative yeast interaction networks, that our algorithm is able to utilize complementary feature sets to make more accurate predictions at the three levels than when the three problems are approached independently. To facilitate application of our multi-level learning framework, we discuss three key aspects of multi-level learning and the corresponding design choices that we have made in the implementation of a concrete learning algorithm. 1) Architecture of information flow: we show the greater flexibility of bidirectional flow over independent levels and unidirectional flow; 2) Coupling mechanism of the different levels: We show how this can be accomplished via augmenting the training sets at each level, and discuss the prevention of error propagation between different levels by means of soft coupling; 3) Sparseness of data: We show that the multi-level framework compounds data sparsity issues, and discuss how this can be dealt with by building local models in information-rich parts of the data. Our proof-of-concept learning algorithm demonstrates the advantage of combining levels, and opens up opportunities for further research. AVAILABILITY: The software and a readme file can be downloaded at http://networks.gersteinlab.org/mll. The programs are written in Java, and can be run on any platform with Java 1.4 or higher and Apache Ant 1.7.0 or higher installed. The software can be used without a license.

UNLABELLED: Residue coevolution has recently emerged as an important concept, especially in the context of protein structures. While a multitude of different functions for quantifying it have been proposed, not much is known about their relative strengths and weaknesses. Also, subtle algorithmic details have discouraged implementing and comparing them. We addressed this issue by developing an integrated online system that enables comparative analyses with a comprehensive set of commonly used scoring functions, including Statistical Coupling Analysis (SCA), Explicit Likelihood of Subset Variation (ELSC), mutual information and correlation-based methods. A set of data preprocessing options are provided for improving the sensitivity and specificity of coevolution signal detection, including sequence weighting, residue grouping and the filtering of sequences, sites and site pairs. A total of more than 100 scoring variations are available. The system also provides facilities for studying the relationship between coevolution scores and inter-residue distances from a crystal structure if provided, which may help in understanding protein structures. AVAILABILITY: The system is available at http://coevolution.gersteinlab.org. The source code and JavaDoc API can also be downloaded from the web site.

PDDL2.1 was designed to push the envelope of what planning algorithms can do, and it has succeeded. It adds two important features: durative actions,which take time (and may have continuous effects); and objective functions for measuring the quality of plans. The concept of durative actions is flawed; and the treatment of their semantics reveals too strong an attachment to the way many contemporary planners work. Future PDDL innovators should focus on producing a clean semantics for additions to the language, and let planner implementers worry about coupling their algorithms to problems expressed in the latest version of the language.

Classical planning is the problem of finding a sequence of actions to achieve a goal given an exact characterization of a domain. An algorithm to solve this problem is presented, which searches a space of plan prefixes, trying to extend one of them to a complete sequence of actions. It is guided by a heuristic estimator based on regression-match graphs, which attempt to characterize the entire subgoal structure of the remaining part of the problem. These graphs simplify the structure by neglecting goal interactions and by assuming that variables in goal conjunctions should be bound in such a way as to make as many conjuncts as possible true without further work. In some domains, these approximations work very well, and experiments show that many classical planning problems can be solved with very little search.

The problem of place recognition is central to robot map learning. A robot needs to be able to recognize when it has returned to a previously visited place, or at least to be able to estimate the likelihood that it has been at a place before. Our approach is to compare images taken at two places, using a stochastic model of changes due to shift, zoom, and occlusion to predict the probability that one of them could be a perturbation of the other. We have performed experiments to gather the valve of a /spl chi//sup 2/ statistic applied to image matching from a variety of indoor locations. Image pairs gathered from nearby locations generate low /spl chi//sup 2/ values, and images gathered from different locations generate high values. The rate of false positive and false negative matches is low.

In this introduction, our main purpose is to survey the field of planning and scheduling, place our collection of papers in that context, and then to point out the field's weak and strong spots and some directions for the future.We'll start with a historical overview.Throughout this introduction, we use this type face when referring to papers in this special issue of Artijicial Intelligence.' "SNLP" stands for "Systematic NonLinear Planner".

Intelligent agency requires some ability to predict the future. An agent must ask itself what is presently its best course of action given what it now knows about what the world will be like when it intends to act. This paper presents a system that uses a probabilistic model to reason about the effects of an agent's proposed actions on a dynamic and uncertain world, computing the probability that relevant propositions will hold at a specified point in time. The model allows for incomplete information about the world, the occurrence of exogenous (unplanned) events, unreliable sensors, and the possibility of an imperfect causal theory. The system provides an application program with answers to questions of the form “is the probability that ϕ will hold in the world at time t greater than τ?” It is unique among algorithms for probabilistic temporal reasoning in that it tries to limit its inference according to the proposition, time, and probability threshold provided by the application. The system will also notify the application if subsequent evidence invalidates its answer to a query.

The Cyc project, and this book, are based on two premises: ( 1 ) that what is holding AI back is programs' lack of explicitly represented knowledge; and (2) that the time is ripe to fix the problem by encoding large chunks of it.The project aims to verify (2) by the most forthright means possible, namely, to carry out the encoding.This book is a report on the progress so far.It does not succeed in convincing me of either (1) or (2).However, it does get the juices flowing; if the time is not yet ripe for building a large knowledge base, the book makes you hope the time is coming soon.Premise (1) is essentially taken for granted by Lenat and Guha. 1 There is a sort of Rip van Winkle quality at work here; they started building Cyc in 1984, and their consciousness got frozen at about that point.As it happens, there is a lot of skepticism about this premise in the AI community, and the authors don't address it.The skepticism stems from two sources

An issue that must be addressed in map-learning systems is that of error accumulation. The primary emphasis in the literature has been on reducing errors entering the map. The authors suggest that this methodology must reach a point of diminishing returns, and hence focus on explicit error detection and correction. By identifying the possible types of mapping errors, structural constraints can be exploited to detect and diagnose mapping errors. Such robust mapping requires little overhead beyond that needed for nonrobust mapping. A mapping system was implemented based on those ideas. Extensive testing in simulation demonstrated the effectiveness of the proposed error-correction strategies.>

There are two strands of research into automated planning: the study of nonlinear, progressive, heuristic planners, and the study of linear, regressive, rigorous planners. We focus on the latter, in particular the formulation of E. Pednault. Regression planning can be viewed as a nondeterministic algorithm in which a goal is reduced by preserving it from the initial situation, by insertion of a step to accomplish it, or by adaptation of an existing step. This algorithm can be proven complete, in the sense that it finds any straight-line plan with no redundant steps. an extension of the algorithm to make use of “formal objects” is also complete. A practical implementation of the algorithm can solve several nontrivial problems. A powerful predicate-calculus simplifier is an important component. It is still unclear whether these elegant results make contact with practicality.

Nonmonotonic formal systems have been proposed as an extension to classical first-order logic that will capture the process of human “default reasoning” or “plausible inference” through their inference mechanisms, just as modus ponens provides a model for deductive reasoning. But although the technical properties of these logics have been studied in detail and many examples of human default reasoning have been identified, for the most part these logics have not actually been applied to practical problems to see whether they produce the expected results. We provide axioms for a simple problem in temporal reasoning which has long been identified as a case of default reasoning, thus presumably amenable to representation in nonmonotonic logic. Upon examining the resulting nonmonotonic theories, however, we find that the inferences permitted by the logics are not those we had intended when we wrote the axioms, and in fact are much weaker. This problem is shown to be independent of the logic used; nor does it depend on any particular temporal representation. Upon analyzing the failure we find that the nonmonotonic logics we considered are inherently incapable of representing this kind of default reasoning. The first part of the paper is an expanded version of one that appeared in the 1986 AAAI proceedings. The second part reports on several responses to our result that have appeared since the original paper was published.

Two data-organization devices that have come out of AI re-search are data pools (``contexts'') and data dependencies. The latter are more flexible than the former, and have supplanted them. Data pools offer certain advantages of efficiency, however, so it is worth trying to make the two mechanisms compatible. Doing this requires generalizing the mark-and-sweep algorithms that maintain consistency in a data-dependency network, so that the labels passed around do not simply say whether a datum is IN or OUT, but say which data pools it is present in. The revised algorithm is essentially an algorithm for solving simultaneous Boolean equations. Other mechanisms are needed for per-forming useful chores like maintaining well-founded support links and orchestrating demon calls.

Tradmonal logics suffer from the "monotomclty problem"' new axioms never mvahdate old theorems One way to get nd of this problem ts to extend traditional modal logic in the following wayThe operator M (usually read "possible") is extended so that Mp is true whenever p is consistent with the theory Then any theorem of this form may be mvahdated if ~p ~s added as an axiom This extension results m nonmonotomc versions of the systems T, $4, and $5 These systems are complete in that a theorem is provable in a theory based on one of them just if it is true m all "noncommittal" models of that theory, where a noncommittal model ts one m which as many thmgs are possible as possible Nonmonotomc $4 is probably the most interesting of the three, since it is stronger than ordinary $4 but has all the usual inferential machinery of $4 There is a straightforward proof procedure for the sententlal subset of nonmonotomc $4.This approach to nonmonotonlc logic may be applied to several problems in knowledge representation for arUficml mtelhgence Its main advantages over competmg approaches are that tt factors out problems of resource hmltattons and allows the symbol M to appear m any context, since M is a meaningful part of the language Categories and Subject Descriptors' F.4 1 ]Mathematical Logic and Formal Languages].Mathematical