TY - GEN
AB - We consider the problem of inference in agraphical model with binary variables. While in theory it is arguably preferable to compute marginal probabilities, in practice researchers often use MAP inference due to the availability of efficient discrete optimization algorithms. We bridge the gap between the two approaches by introducing the Discrete Marginals technique in which approximate marginals are obtained by minimizing an objective function with unary and pair-wise terms over a discretized domain. This allows the use of techniques originally devel-oped for MAP-MRF inference and learning. We explore two ways to set up the objective function - by discretizing the Bethe free energy and by learning it from training data. Experimental results show that for certain types of graphs a learned function can out-perform the Bethe approximation. We also establish a link between the Bethe free energy and submodular functions.
AU - Korc, Filip
AU - Kolmogorov, Vladimir
AU - Lampert, Christoph
ID - 5396
SN - 2664-1690
TI - Approximating marginals using discrete energy minimization
ER -
TY - GEN
AB - This document is created as a part of the project “Repository for Research Data on IST Austria”. It summarises the actual state of research data at IST Austria, based on survey results. It supports the choice of appropriate software, which would best fit the requirements of their users, the researchers.
AU - Porsche, Jana
ID - 5398
TI - Actual state of research data @ ISTAustria
ER -
TY - CHAP
AU - Gupta, Ashutosh
ID - 5745
SN - 0302-9743
T2 - Automated Technology for Verification and Analysis
TI - Improved Single Pass Algorithms for Resolution Proof Reduction
VL - 7561
ER -
TY - CONF
AB - We consider Markov decision processes (MDPs) with specifications given as Büchi (liveness) objectives. We consider the problem of computing the set of almost-sure winning vertices from where the objective can be ensured with probability 1. We study for the first time the average case complexity of the classical algorithm for computing the set of almost-sure winning vertices for MDPs with Büchi objectives. Our contributions are as follows: First, we show that for MDPs with constant out-degree the expected number of iterations is at most logarithmic and the average case running time is linear (as compared to the worst case linear number of iterations and quadratic time complexity). Second, for the average case analysis over all MDPs we show that the expected number of iterations is constant and the average case running time is linear (again as compared to the worst case linear number of iterations and quadratic time complexity). Finally we also show that given that all MDPs are equally likely, the probability that the classical algorithm requires more than constant number of iterations is exponentially small.
AU - Chatterjee, Krishnendu
AU - Joglekar, Manas
AU - Shah, Nisarg
ID - 2715
TI - Average case analysis of the classical algorithm for Markov decision processes with Büchi objectives
VL - 18
ER -
TY - CONF
AB - We study the problem of maximum marginal prediction (MMP) in probabilistic graphical models, a task that occurs, for example, as the Bayes optimal decision rule under a Hamming loss. MMP is typically performed as a two-stage procedure: one estimates each variable's marginal probability and then forms a prediction from the states of maximal probability. In this work we propose a simple yet effective technique for accelerating MMP when inference is sampling-based: instead of the above two-stage procedure we directly estimate the posterior probability of each decision variable. This allows us to identify the point of time when we are sufficiently certain about any individual decision. Whenever this is the case, we dynamically prune the variables we are confident about from the underlying factor graph. Consequently, at any time only samples of variables whose decision is still uncertain need to be created. Experiments in two prototypical scenarios, multi-label classification and image inpainting, show that adaptive sampling can drastically accelerate MMP without sacrificing prediction accuracy.
AU - Lampert, Christoph
ID - 2825
TI - Dynamic pruning of factor graphs for maximum marginal prediction
VL - 1
ER -
TY - JOUR
AB - We study evolutionary game theory in a setting where individuals learn from each other. We extend the traditional approach by assuming that a population contains individuals with different learning abilities. In particular, we explore the situation where individuals have different search spaces, when attempting to learn the strategies of others. The search space of an individual specifies the set of strategies learnable by that individual. The search space is genetically given and does not change under social evolutionary dynamics. We introduce a general framework and study a specific example in the context of direct reciprocity. For this example, we obtain the counter intuitive result that cooperation can only evolve for intermediate benefit-to-cost ratios, while small and large benefit-to-cost ratios favor defection. Our paper is a step toward making a connection between computational learning theory and evolutionary game dynamics.
AU - Chatterjee, Krishnendu
AU - Zufferey, Damien
AU - Nowak, Martin
ID - 2848
JF - Journal of Theoretical Biology
TI - Evolutionary game dynamics in populations with different learners
VL - 301
ER -
TY - JOUR
AU - Edelsbrunner, Herbert
AU - Strelkova, Nataliya
ID - 2849
IS - 6
JF - Russian Mathematical Surveys
TI - On the configuration space of Steiner minimal trees
VL - 67
ER -
TY - CONF
AB - Formal verification aims to improve the quality of hardware and software by detecting errors before they do harm. At the basis of formal verification lies the logical notion of correctness, which purports to capture whether or not a circuit or program behaves as desired. We suggest that the boolean partition into correct and incorrect systems falls short of the practical need to assess the behavior of hardware and software in a more nuanced fashion against multiple criteria.
AU - Henzinger, Thomas A
ID - 2888
T2 - Conference proceedings MODELS 2012
TI - Quantitative reactive models
VL - 7590
ER -
TY - CONF
AB - Systems are often specified using multiple requirements on their behavior. In practice, these requirements can be contradictory. The classical approach to specification, verification, and synthesis demands more detailed specifications that resolve any contradictions in the requirements. These detailed specifications are usually large, cumbersome, and hard to maintain or modify. In contrast, quantitative frameworks allow the formalization of the intuitive idea that what is desired is an implementation that comes "closest" to satisfying the mutually incompatible requirements, according to a measure of fit that can be defined by the requirements engineer. One flexible framework for quantifying how "well" an implementation satisfies a specification is offered by simulation distances that are parameterized by an error model. We introduce this framework, study its properties, and provide an algorithmic solution for the following quantitative synthesis question: given two (or more) behavioral requirements specified by possibly incompatible finite-state machines, and an error model, find the finite-state implementation that minimizes the maximal simulation distance to the given requirements. Furthermore, we generalize the framework to handle infinite alphabets (for example, realvalued domains). We also demonstrate how quantitative specifications based on simulation distances might lead to smaller and easier to modify specifications. Finally, we illustrate our approach using case studies on error correcting codes and scheduler synthesis.
AU - Cerny, Pavol
AU - Gopi, Sivakanth
AU - Henzinger, Thomas A
AU - Radhakrishna, Arjun
AU - Totla, Nishant
ID - 2890
T2 - Proceedings of the tenth ACM international conference on Embedded software
TI - Synthesis from incompatible specifications
ER -
TY - CONF
AB - Quantitative automata are nondeterministic finite automata with edge weights. They value a
run by some function from the sequence of visited weights to the reals, and value a word by its
minimal/maximal run. They generalize boolean automata, and have gained much attention in
recent years. Unfortunately, important automaton classes, such as sum, discounted-sum, and
limit-average automata, cannot be determinized. Yet, the quantitative setting provides the potential
of approximate determinization. We define approximate determinization with respect to
a distance function, and investigate this potential.
We show that sum automata cannot be determinized approximately with respect to any
distance function. However, restricting to nonnegative weights allows for approximate determinization
with respect to some distance functions.
Discounted-sum automata allow for approximate determinization, as the influence of a word’s
suffix is decaying. However, the naive approach, of unfolding the automaton computations up
to a sufficient level, is shown to be doubly exponential in the discount factor. We provide an
alternative construction that is singly exponential in the discount factor, in the precision, and
in the number of states. We prove matching lower bounds, showing exponential dependency on
each of these three parameters.
Average and limit-average automata are shown to prohibit approximate determinization with
respect to any distance function, and this is the case even for two weights, 0 and 1.
AU - Boker, Udi
AU - Henzinger, Thomas A
ID - 2891
T2 - Leibniz International Proceedings in Informatics
TI - Approximate determinization of quantitative automata
VL - 18
ER -