The Fourier representation for the uniform distribution over the Boolean cube has found numerous applications in algorithms and complexity analysis. Notably, in learning theory, the learnability of Disjunctive Normal Form (DNF) under the uniform and product distributions has been established through such representations. This paper makes three main contributions. First, it introduces a generalized Fourier expansion that can be used with any distribution $D$ through the representation of the distribution as a Bayesian network (BN). Second, it shows that the main algorithmic tools for learning with the Fourier representation that use membership queries to approximate functions by recovering their heavy Fourier coefficients, can be used with slight modifications with the generalized expansion. These results hold for any distribution. Third, it analyzes the $L_1$ spectral norm of conjunctions under the new expansion, showing that it is bounded for a class of distributions which can be represented by a difference-bounded tree BN, where a parent node in the BN representation can change the conditional expectation of a child node by at most $\alpha<0.5$. Lower bounds are presented to show that such constraints are necessary. Combining these contributions, the paper shows learnability of DNF with membership queries under difference-bounded tree BN.

Continuous time systems are often modeled using discrete time dynamics but this requires a small simulation step to maintain accuracy. In turn, this requires a large planning horizon which leads to computationally demanding planning problems and reduced performance. Previous work in model-free reinforcement learning has partially addressed this issue using action repeats where a policy is learned to determine a discrete action duration. Instead we propose to control the continuous decision timescale directly by using temporally-extended actions and letting the planner treat the duration of the action as an additional optimization variable along with the standard action variables. This additional structure has multiple advantages. It speeds up simulation time of trajectories and, importantly, it allows for deep horizon search in terms of primitive actions while using a shallow search depth in the planner. In addition, in the model-based reinforcement learning (MBRL) setting, it reduces compounding errors from model learning and improves training time for models. We show that this idea is effective and that the range for action durations can be automatically selected using a multi-armed bandit formulation and integrated into the MBRL framework. An extensive experimental evaluation both in planning and in MBRL, shows that our approach yields faster planning, better solutions, and that it enables solutions to problems that are not solved in the standard formulation.

The paper introduces DiSProD, an online planner developed for environments with probabilistic transitions in continuous state and action spaces. DiSProD builds a symbolic graph that captures the distribution of future trajectories, conditioned on a given policy, using independence assumptions and approximate propagation of distributions. The symbolic graph provides a differentiable representation of the policy's value, enabling efficient gradient-based optimization for long-horizon search. The propagation of approximate distributions can be seen as an aggregation of many trajectories, making it well-suited for dealing with sparse rewards and stochastic environments. An extensive experimental evaluation compares DiSProD to state-of-the-art planners in discrete-time planning and real-time control of robotic systems. The proposed method improves over existing planners in handling stochastic environments, sensitivity to search depth, sparsity of rewards, and large action spaces. Additional real-world experiments demonstrate that DiSProD can control ground vehicles and surface vessels to successfully navigate around obstacles.

The paper provides a thorough investigation of Direct Loss Minimization (DLM), which optimizes the posterior to minimize predictive loss, in sparse Gaussian processes. For the conjugate case, we consider DLM for log-loss and DLM for square loss showing a significant performance improvement in both cases. The application of DLM in non-conjugate cases is more complex because the logarithm of expectation in the log-loss DLM objective is often intractable and simple sampling leads to biased estimates of gradients. The paper makes two technical contributions to address this. First, a new method using product sampling is proposed, which gives unbiased estimates of gradients (uPS) for the objective function. Second, a theoretical analysis of biased Monte Carlo estimates (bMC) shows that stochastic gradient descent converges despite the biased gradients. Experiments demonstrate empirical success of DLM. A comparison of the sampling methods shows that, while uPS is potentially more sample-efficient, bMC provides a better tradeoff in terms of convergence time and computational efficiency.

The paper introduces a new algorithm for planning in partially observable Markov decision processes (POMDP) based on the idea of aggregate simulation. The algorithm uses product distributions to approximate the belief state and shows how to build a representation graph of an approximate action-value function over belief space. The graph captures the result of simulating the model in aggregate under independence assumptions, giving a symbolic representation of the value function. The algorithm supports large observation spaces using sampling networks, a representation of the process of sampling values of observations, which is integrated into the graph representation. Following previous work in MDPs this approach enables action selection in POMDPs through gradient optimization over the graph representation. This approach complements recent algorithms for POMDPs which are based on particle representations of belief states and an explicit search for action selection. Our approach enables scaling to large factored action spaces in addition to large state spaces and observation spaces. An experimental evaluation demonstrates that the algorithm provides excellent performance relative to state of the art in large POMDP problems.

It is well known that the problems of stochastic planning and probabilistic inference are closely related. This paper makes two contributions in this context. The first is to provide an analysis of the recently developed SOGBOFA heuristic planning algorithm that was shown to be effective for problems with large factored state and action spaces. It is shown that SOGBOFA can be seen as a specialized inference algorithm that computes its solutions through a combination of a symbolic variant of belief propagation and gradient ascent. The second contribution is a new solver for Marginal MAP (MMAP) inference. We introduce a new reduction from MMAP to maximum expected utility problems which are suitable for the symbolic computation in SOGBOFA. This yields a novel algebraic gradient-based solver (AGS) for MMAP. An experimental evaluation illustrates the potential of AGS in solving difficult MMAP problems.

Hybrid (mixed discrete and continuous) state and action Markov Decision Processes (HSA-MDPs) provide an expressive formalism for modeling stochastic and concurrent sequential decision-making problems. Existing solvers for HSA-MDPs are either limited to very restricted transition distributions, require knowledge of domain-specific basis functions to achieve good approximations, or do not scale. We explore a domain-independent approach based on the framework of hindsight optimization (HOP) for HSA-MDPs, which uses an upper bound on the finite-horizon action values for action selection. Our main contribution is a linear time reduction to a Mixed Integer Linear Program (MILP) that encodes the HOP objective, when the dynamics are specified as location-scale probability distributions parametrized by Piecewise Linear (PWL) functions of states and actions. In addition, we show how to use the same machinery to select actions based on a lower-bound generated by straight line plans. Our empirical results show that the HSA-HOP approach effectively scales to high-dimensional problems and outperforms baselines that are capable of scaling to such large hybrid MDPs.

Bayesian models are established as one of the main successful paradigms for complex problems in machine learning. To handle intractable inference, research in this area has developed new approximation methods that are fast and effective. However, theoretical analysis of the performance of such approximations is not well developed. The paper furthers such analysis by providing bounds on the excess risk of variational inference algorithms and related regularized loss minimization algorithms for a large class of latent variable models with Gaussian latent variables. We strengthen previous results for variational algorithms by showing they are competitive with any point-estimate predictor. Unlike previous work, we also provide bounds on the risk of the \emph{Bayesian} predictor and not just the risk of the Gibbs predictor for the same approximate posterior. The bounds are applied in complex models including sparse Gaussian processes and correlated topic models. Theoretical results are complemented by identifying novel approximations to the Bayesian objective that attempt to minimize the risk directly. An empirical evaluation compares the variational and new algorithms shedding further light on their performance.

Latent Gaussian Models (LGM) provide a rich modeling framework with general inference procedures. The variational approximation offers an effective solution for such models and has attracted a significant amount of interest. Recent work proposed a fixed-point (FP) update procedure to optimize the covariance matrix in the variational solution and demonstrated its efficacy in specific models. The paper makes three contributions. First, it shows that the same approach can be used more generally in extensions of LGM. Second, it provides an analysis identifying conditions for the convergence of the FP method. Third, it provides an extensive experimental evaluation in Gaussian processes, sparse Gaussian processes, and generalized linear models, with several non-conjugate observation likelihoods, showing wide applicability of the FP method and a significant advantage over gradient based optimization.

This paper investigates stochastic planning problemswith large factored state and action spaces. We show that even with moderate increase in the size of existing challenge problems, the performance of state of the art algorithms deteriorates rapidly, making them ineffective.To address this problem we propose a family of simple but scalable online planning algorithms that combine sampling, as in Monte Carlo tree search, with “aggregation,” where the aggregation approximates a distribution over random variables by the product of their marginals. The algorithms are correct under some rather strong technical conditions and can serve as an unsound but effective heuristic when the conditions do not hold. An extensive experimental evaluation demonstrates that the new algorithms provide significant improvement over the state of the art when solving largeproblems in a number of challenge benchmark domains.

Gaussian processes (GP) provide an attractive machine learning model due to their non-parametric form, their flexibility to capture many types of observation data, and their generic inference procedures. Sparse GP inference algorithms address the cubic complexity of GPs by focusing on a small set of pseudo-samples. To date, such approaches have focused on the simple case of Gaussian observation likelihoods. This paper develops a variational sparse solution for GPs under general likelihoods by providing a new characterization of the gradients required for inference in terms of individual observation likelihood terms. In addition, we propose a simple new approach for optimizing the sparse variational approximation using a fixed point computation. We demonstrate experimentally that the fixed point operator acts as a contraction in many cases and therefore leads to fast convergence. An experimental evaluation for count regression, classification, and ordinal regression illustrates the generality and advantages of the new approach.

Recent work introduced Generalized First Order Decision Diagrams (GFODD) as a knowledge representation that is useful in mechanizing decision theoretic planning in relational domains. GFODDs generalize function-free first order logic and include numerical values and numerical generalizations of existential and universal quantification. Previous work presented heuristic inference algorithms for GFODDs. In this paper, we study the complexity of the evaluation problem, the satiability problem, and the equivalence problem for GFODDs under the assumption that the size of the intended model is given with the problem, a restriction that guarantees decidability. Our results provide a complete characterization. The same characterization applies to the corresponding restriction of problems in first order logic, giving an interesting new avenue for efficient inference when the number of objects is bounded. Our results show that for Σk formulas, and for corresponding GFODDs, evaluation and satisfiability are Σkp complete, and equivalence is Πk+1p complete. For Πk formulas evaluation is Πkp complete, satisfiability is one level higher and is Σk+1p complete, and equivalence is Πk+1p complete.

We address the scalability of symbolic planning under uncertainty with factored states and actions. Prior work has focused almost exclusively on factored states but not factored actions, and on value iteration (VI) compared to policy iteration (PI). Our first contribution is a novel method for symbolic policy backups via the application of constraints, which is used to yield a new efficient symbolic imple- mentation of modified PI (MPI) for factored action spaces. While this approach improves scalability in some cases, naive handling of policy constraints comes with its own scalability issues. This leads to our second and main contribution, symbolic Opportunistic Policy Iteration (OPI), which is a novel convergent al- gorithm lying between VI and MPI. The core idea is a symbolic procedure that applies policy constraints only when they reduce the space and time complexity of the update, and otherwise performs full Bellman backups, thus automatically adjusting the backup per state. We also give a memory bounded version of this algorithm allowing a space-time tradeoff. Empirical results show significantly improved scalability over the state-of-the-art.

We consider symbolic dynamic programming (SDP) for solving Markov Decision Processes (MDP) with factored state and action spaces, where both states and actions are described by sets of discrete variables. Prior work on SDP has considered only the case of factored states and ignored structure in the action space, causing them to scale poorly in terms of the number of action variables. Our main contribution is to present the first SDP-based planning algorithm for leveraging both state and action space structure in order to compute compactly represented value functions and policies. Since our new algorithm can potentially require more space than when action structure is ignored, our second contribution is to describe an approach for smoothly trading-off space versus time via recursive conditioning. Finally, our third contribution is to introduce a novel SDP approximation that often significantly reduces planning time with little loss in quality by exploiting action structure in weakly coupled MDPs. We present empirical results in three domains with factored action spaces that show that our algorithms scale much better with the number of action variables as compared to state-of-the-art SDP algorithms.

Hybrid reactive-deliberative architectures in robotics combine reactive sub-policies for fast action execution with goal sequencing and deliberation. The need for replanning, however, presents a challenge for reactivity and hinders the potential for guarantees about the plan quality. In this paper, we argue that one can integrate abstract planning provided by symbolic dynamic programming in first order logic into a reactive robotic architecture, and that such an integration is in fact natural and has advantages over traditional approaches. In particular, it allows the integrated system to spend off-line time planning for a policy, and then use the policy reactively in open worlds, in situations with unexpected outcomes, and even in new environments, all by simply reacting to a state change executing a new action proposed by the policy. We demonstrate the viability of the approach by integrating the FODD-Planner with the robotic DIARC architecture showing how an appropriate interface can be defined and that this integration can yield robust goal-based action execution on robots in open worlds.

Large-scale contamination of ground water due to improper disposal of hazardous chemicals poses a global threat to drinking water supplies. Effective restoration and remediation of such sites relies upon a knowledge of the contaminant's distribution within the subsurface. Obtaining a detailed map of the existing distribution is usually not feasible; rather partial knowledge in terms of certain metrics that characterize the distribution has recently been shown to be sufficient for planning and monitoring remediation strategies. In this work we explore the prediction of a representative metric based upon down-gradient concentration profiles using a classification framework where each class represents a particular sub-range of the metric. Initial experiments show that our proposed model can be used effectively for predicting the metric.

Dynamic programming algorithms have been successfully applied to propositional stochastic planning problems by using compact representations, in particular algebraic decision diagrams, to capture domain dynamics and value functions. Work on symbolic dynamic programming lifted these ideas to first order logic using several representation schemes. Recent work introduced a first order variant of decision diagrams (FODD) and developed a value iteration algorithm for this representation. This paper develops several improvements to the FODD algorithm that make the approach practical. These include, new reduction operators that decrease the size of the representation, several speedup techniques, and techniques for value approximation. Incorporating these, the paper presents a planning system, FODD-Planner, for solving relational stochastic planning problems. The system is evaluated on several domains, including problems from the recent international planning competition, and shows competitive performance with top ranking systems. This is the first demonstration of feasibility of this approach and it shows that abstraction through compact representation is a promising approach to stochastic planning.

Relational Markov Decision Processes (MDP) are a useful abstraction for stochastic planning problems since one can develop abstract solutions for them that are independent of domain size or instantiation. While there has been an increased interest in developing relational fully observable MDPs, there has been very little work on relational partially observable MDPs (POMDP), which deal with uncertainty in problem states in addition to stochastic action effects. This paper provides a concrete formalization of relational POMDPs making several technical contributions toward their solution. First, we show that to maintain correctness one must distinguish between quantification over states and quantification over belief states; this implies that solutions based on value iteration are inherently limited to the finite horizon case. Second, we provide a symbolic dynamic programing algorithm for finite horizon relational POMDPs, solving them at an abstract level, by lifting the propositional incremental pruning algorithm. Third, we show that this algorithm can be implemented using first order decision diagrams, a compact representation for functions over relational structures, that has been recently used to solve relational MDPs.

Two aspects are crucial when constructing any real world supervised classification task: the set of classes whose distinction might be useful for the domain expert, and the set of classifications that can actually be distinguished by the data. Often a set of labels is defined with some initial intuition but these are not the best match for the task. For example, labels have been assigned for land cover classification of the Earth but it has been suspected that these labels are not ideal and some classes may be best split into subclasses whereas others should be merged. This paper formalizes this problem using three ingredients: the existing class labels, the underlying separability in the data, and a special type of input from the domain expert. We require a domain expert to specify an L × L matrix of pairwise probabilistic constraints expressing their beliefs as to whether the L classes should be kept separate, merged, or split. This type of input is intuitive and easy for experts to supply. We then show that the problem can be solved by casting it as an instance of penalized probabilistic clustering (PPC). Our method, Class-Level PPC (CPPC) extends PPC showing how its time complexity can be reduced from O(N2) to O(NL) for the problem of class re-definition. We further extend the algorithm by presenting a heuristic to measure adherence to constraints, and providing a criterion for determining the model complexity (number of classes) for constraint-based clustering. We demonstrate and evaluate CPPC on artificial data and on our motivating domain of land cover classification. For the latter, an evaluation by domain experts shows that the algorithm discovers novel class definitions that are better suited to land cover classification than the original set of labels.

Markov decision processes capture sequential decision making under uncertainty, where an agent must choose actions so as to optimize long term reward. The paper studies efficient reasoning mechanisms for Relational Markov Decision Processes (RMDP) where world states have an internal relational structure that can be naturally described in terms of objects and relations among them. Two contributions are presented. First, the paper develops First Order Decision Diagrams (FODD), a new compact representation for functions over relational structures, together with a set of operators to combine FODDs, and novel reduction techniques to keep the representation small. Second, the paper shows how FODDs can be used to develop solutions for RMDPs, where reasoning is performed at the abstract level and the resulting optimal policy is independent of domain size (number of objects) or instantiation. In particular, a variant of the value iteration algorithm is developed by using special operations over FODDs, and the algorithm is shown to converge to the optimal policy.

The paper presents a kernel for learning from ordered hypergraphs, a formalization that captures relational data as used in Inductive Logic Programming (ILP). The kernel generalizes previous approaches to graph kernels in calculating similarity based on walks in the hypergraph. Experiments on challenging chemical datasets demonstrate that the kernel outperforms existing ILP methods, and is competitive with state-of-the-art graph kernels. The experiments also demonstrate that the encoding of graph data can affect performance dramatically, a fact that can be useful beyond kernel methods.

A large number of variants of the Perceptron algorithm have been proposed and partially evaluated in recent work. One type of algorithm aims for noise tolerance by replacing the last hypothesis of the perceptron with another hypothesis or a vote among hypotheses. Another type simply adds a margin term to the perceptron in order to increase robustness and accuracy, as done in support vector machines. A third type borrows further from support vector machines and constrains the update function of the perceptron in ways that mimic soft-margin techniques. The performance of these algorithms, and the potential for combining different techniques, has not been studied in depth. This paper provides such an experimental study and reveals some interesting facts about the algorithms. In particular the perceptron with margin is an effective method for tolerating noise and stabilizing the algorithm. This is surprising since the margin in itself is not designed or used for noise tolerance, and there are no known guarantees for such performance. In most cases, similar performance is obtained by the voted-perceptron which has the advantage that it does not require parameter selection. Techniques using soft margin ideas are run-time intensive and do not give additional performance benefits. The results also highlight the difficulty with automatic parameter selection which is required with some of these variants.

The paper studies machine learning problems where each example is described using a set of Boolean features and where hypotheses are represented by linear threshold elements. One method of increasing the expressiveness of learned hypotheses in this context is to expand the feature set to include conjunctions of basic features. This can be done explicitly or where possible by using a kernel function. Focusing on the well known Perceptron and Winnow algorithms, the paper demonstrates a tradeoff between the computational efficiency with which the algorithm can be run over the expanded feature space and the generalization ability of the corresponding learning algorithm. We first describe several kernel functions which capture either limited forms of conjunctions or all conjunctions. We show that these kernels can be used to efficiently run the Perceptron algorithm over a feature space of exponentially many conjunctions; however we also show that using such kernels, the Perceptron algorithm can provably make an exponential number of mistakes even when learning simple functions. We then consider the question of whether kernel functions can analogously be used to run the multiplicative-update Winnow algorithm over an expanded feature space of exponentially many conjunctions. Known upper bounds imply that the Winnow algorithm can learn Disjunctive Normal Form (DNF) formulae with a polynomial mistake bound in this setting. However, we prove that it is computationally hard to simulate Winnow's behavior for learning DNF over such a feature set. This implies that the kernel functions which correspond to running Winnow for this problem are not efficiently computable, and that there is no general construction that can run Winnow with kernels.

Recent work has introduced Boolean kernels with which one can learn linear threshold functions over a feature space containing all conjunctions of length up to k (for any 1 ≤ k ≤ n) over the original n Boolean features in the input space. This motivates the question of whether maximum margin algorithms such as Support Vector Machines can learn Disjunctive Normal Form expressions in the Probably Approximately Correct (PAC) learning model by using this kernel. We study this question, as well as a variant in which structural risk minimization (SRM) is performed where the class hierarchy is taken over the length of conjunctions. We show that maximum margin algorithms using the Boolean kernels do not PAC learn t(n)-term DNF for any t(n) = ω(1), even when used with such a SRM scheme. We also consider PAC learning under the uniform distribution and show that if the kernel uses conjunctions of length ˜ω(√n) then the maximum margin hypothesis will fail on the uniform distribution as well. Our results concretely illustrate that margin based algorithms may overfit when learning simple target functions with natural kernels.

Data mining can be viewed, in many instances, as the task of computing a representation of a theory of a model or a database, in particular by finding a set of maximally specific sentences satisfying some property. We prove some hardness results that rule out simple approaches to solving the problem.The a priori algorithm is an algorithm that has been successfully applied to many instances of the problem. We analyze this algorithm, and prove that is optimal when the maximally specific sentences are "small". We also point out its limitations.We then present a new algorithm, the Dualize and Advance algorithm, and prove worst-case complexity bounds that are favorable in the general case. Our results use the concept of hypergraph transversals. Our analysis shows that the a priori algorithm can solve the problem of enumerating the transversals of a hypergraph, improving on previously known results in a special case. On the other hand, using results for the general case of the hypergraph transversal enumeration problem, we can show that the Dualize and Advance algorithm has worst-case running time that is sub-exponential to the output size (i.e., the number of maximally specific sentences).We further show that the problem of finding maximally specific sentences is closely related to the problem of exact learning with membership queries studied in computational learning theory.

We study online learning in Boolean domains using kernels which cap- ture feature expansions equivalent to using conjunctions over basic fea- tures. We demonstrate a tradeoff between the computational efficiency with which these kernels can be computed and the generalization abil- ity of the resulting classifier. We first describe several kernel functions which capture either limited forms of conjunctions or all conjunctions. We show that these kernels can be used to efficiently run the Percep- tron algorithm over an exponential number of conjunctions; however we also prove that using such kernels the Perceptron algorithm can make an exponential number of mistakes even when learning simple func- tions. We also consider an analogous use of kernel functions to run the multiplicative-update Winnow algorithm over an expanded feature space of exponentially many conjunctions. While known upper bounds imply that Winnow can learn DNF formulae with a polynomial mistake bound in this setting, we prove that it is computationally hard to simulate Win- now’s behavior for learning DNF over such a feature set, and thus that such kernel functions for Winnow are not efficiently computable.

We study the learnability of read-k-satisfy-j (RkSj) DNF formulas. These are boolean formulas in disjunctive normal form (DNF), in which the maximum number of occurrences of a variable is bounded by k, and the number of terms satisfied by any assignment is at most j. After motivating the investigation of this class of DNF formulas, we present an algorithm that for any unknown RkSj DNF formula to be learned, with high probability finds a logically equivalent DNF formula using the well-studied protocol of equivalence and membership queries. The algorithm runs in polynomial time for $k\cdot j=O({\log n\over\log\log n})$, where n is the number of input variables.

Several data mining problems can be formulated as problems of finding maximally specific sentences that are interesting in a database. We first show that this problem has a close relationship with the hypergraph transversal problem. We then analyze two algorithms that have been previously used in data mining, proving upper bounds on their complexity. The first algorithm is useful when the maximally specific interesting sentences are "small". We show that this algorithm can also be used to efficiently solve a special case of the hypergraph transversal problem, improving on previous results. The second algorithm utilizes a subroutine for hypergraph transversals, and is applicable in more general situations, with complexity close to a lower bound for the problem. We also relate these problems to the model of exact learning in computational learning theory, and use the correspondence to derive some corollaries.

We introduce a new framework for the study of reasoning. The Learning (in order) to Reason approach developed here views learning as an integral part of the inference process, and suggests that learning and reasoning should be studied together. The Learning to Reason framework combines the interfaces to the world used by known learning models with the reasoning task and a performance criterion suitable for it. In this framework, the intelligent agent is given access to its favorite learning interface, and is also given a grace period in with it can interact with this interface and construct a representation KB of the world W . The reasoning performance is measured only after this period, when the agent is presented with queries α from some query language, relevant to the world, and has to answer whether W implies α. The approach is meant to overcome the main computational difficulties in the traditional treatment of reasoning which stem from its separation from the “world”. Since the agent interacts with the world when construction its knowledge representation it can choose a representation that is useful for the task at hand. Moreover, we can now make explicit the dependence of the reasoning performance on the environment the agent interacts with. We show how previous results from learning theory and reasoning fit into this framwork and illustrate the usefulness of the Learning to Reason approach by exhibiting new results that are not possible in the traditional setting. First, we give Learning to Reason algorithms for classes of propositional languages for which there are no efficient reasoning algorithms, when represented as a traditional (formula-based) knowledge base. Second, we exhibit a Learning to Reason algorithm for a class of propositional languages that is not know to be learnable in the traditional sense.

The Learning to Reason framework combines the study of Learning and Reasoning into a single task. Within it, learning is done specifically for the purpose of reasoning with the learned knowledge. Computational considerations show that this is a useful paradigm; in some cases learning and reasoning problems that are intractable when studied separately become tractable when performed as a task of Learning to Reason. In this paper we study Learning to Reason problems where the interaction with the world supplies the learner only partial information in the form of partial assignments. Several natural interpretations of partial assignments are considered and learning and reasoning algorithms using these are developed. The results presented exhibit a tradeo between learnability, the strength of the oracles used in the interface, and the range of reasoning queries the learner is guaranteed to answer correctly.

Characteristic models are an alternative, model based, representation for Horn expressions. It has been shown that these two representations are incomparable and each has its advantages over the other. It is therefore natural to ask what is the cost of translating, back and forth, between these representations. Interestingly, the same translation questions arise in database theory, where it has applications to the design of relational databases. This paper studies the computational complexity of these problems. Our main result is that the two translation problems are equivalent under polynomial reductions, and that they are equivalent to the corresponding decision problem. Namely, translating is equivalent to deciding whether a given set of models is the set of characteristic models for a given Horn expression. We also relate these problems to the hypergraph transversal problem, a well known problem which is related to other applications in AI and for which no polynomial time algorithm is known. It is shown that in general our translation problems are at least as hard as the hypergraph transversal problem, and in a special case they are equivalent to it.

We study the learnability of Read-k-Satisfy-j (RkSj) DNF formulae. These are DNF formulae in which the maximal number of occurrences of a variable is bounded by k, and the number of terms satisfied by any assignment is at most j. We show that this class of functions is learnable in polynomial time, using Equivalence and Membership Queries, as long as k•j=O(logn/loglogn). Learnability was previously known only in case that both k and j are constants. We also present a family of boolean functions that have short (poly(n)) Read-2-Satisfy-1 DNF formulae but require CNF formulae of size > 2W(n). Therefore, our result does not seem to follow from the recent learnability result of [Bsh93].