Interval Markov Decision Processes (IMDPs) are finite-state uncertain Markov models, where the transition probabilities belong to intervals. Recently, there has been a surge of research on employing IMDPs as abstractions of stochastic systems for control synthesis. However, due to the absence of algorithms for synthesis over IMDPs with continuous action-spaces, the action-space is assumed discrete a-priori, which is a restrictive assumption for many applications. Motivated by this, we introduce continuous-action IMDPs (caIMDPs), where the bounds on transition probabilities are functions of the action variables, and study value iteration for maximizing expected cumulative rewards. Specifically, we decompose the max-min problem associated to value iteration to $|\mathcal{Q}|$ max problems, where $|\mathcal{Q}|$ is the number of states of the caIMDP. Then, exploiting the simple form of these max problems, we identify cases where value iteration over caIMDPs can be solved efficiently (e.g., with linear or convex programming). We also gain other interesting insights: e.g., in certain cases where the action set $\mathcal{A}$ is a polytope, synthesis over a discrete-action IMDP, where the actions are the vertices of $\mathcal{A}$, is sufficient for optimality. We demonstrate our results on a numerical example. Finally, we include a short discussion on employing caIMDPs as abstractions for control synthesis.
相關內容
In this paper, we consider point sets of finite Desarguesian planes whose multisets of intersection numbers with lines is the same for all but one exceptional parallel class of lines. We call such sets regular of affine type. When the lines of the exceptional parallel class have the same intersection numbers, then we call these sets regular of pointed type. Classical examples are e.g. unitals; a detailed study and constructions of such sets with few intersection numbers is due to Hirschfeld and Sz\H{o}nyi from 1991. We here provide some general construction methods for regular sets and describe a few infinite families. The members of one of these families have the size of a unital and meet affine lines of $\mathrm{PG}(2, q^2)$ in one of $4$ possible intersection numbers, each of them congruent to $1$ modulo $\sqrt{q}$. As a byproduct, we determine the intersection sizes of the Hermitian curve defined over $\mathrm{GF}(q^2)$ with suitable rational curves of degree $\sqrt{q}$ and we obtain $\sqrt{q}$-divisible codes with $5$ non-zero weights. We also determine the weight enumerator of the codes arising from the general constructions modulus some $q$-powers.
Multivariate temporal, or time, series classification is, in a way, the temporal generalization of (numeric) classification, as every instance is described by multiple time series instead of multiple values. Symbolic classification is the machine learning strategy to extract explicit knowledge from a data set, and the problem of symbolic classification of multivariate temporal series requires the design, implementation, and test of ad-hoc machine learning algorithms, such as, for example, algorithms for the extraction of temporal versions of decision trees. One of the most well-known algorithms for decision tree extraction from categorical data is Quinlan's ID3, which was later extended to deal with numerical attributes, resulting in an algorithm known as C4.5, and implemented in many open-sources data mining libraries, including the so-called Weka, which features an implementation of C4.5 called J48. ID3 was recently generalized to deal with temporal data in form of timelines, which can be seen as discrete (categorical) versions of multivariate time series, and such a generalization, based on the interval temporal logic HS, is known as Temporal ID3. In this paper we introduce Temporal C4.5, that allows the extraction of temporal decision trees from undiscretized multivariate time series, describe its implementation, called Temporal J48, and discuss the outcome of a set of experiments with the latter on a collection of public data sets, comparing the results with those obtained by other, classical, multivariate time series classification methods.
In this paper we study online change point detection in dynamic networks with time heterogeneous missing pattern within networks and dependence across the time course. The missingness probabilities, the entrywise sparsity of networks, the rank of networks and the jump size in terms of the Frobenius norm, are all allowed to vary as functions of the pre-change sample size. On top of a thorough handling of all the model parameters, we notably allow the edges and missingness to be dependent. To the best of our knowledge, such general framework has not been rigorously nor systematically studied before in the literature. We propose a polynomial time change point detection algorithm, with a version of soft-impute algorithm (e.g. Mazumder et al., 2010; Klopp, 2015) as the imputation sub-routine. Piecing up these standard sub-routines algorithms, we are able to solve a brand new problem with sharp detection delay subject to an overall Type-I error control. Extensive numerical experiments are conducted demonstrating the outstanding performances of our proposed method in practice.
Deep neural networks (DNNs) trained to minimize a loss term plus the sum of squared weights via gradient descent corresponds to the common approach of training with weight decay. This paper provides new insights into this common learning framework. We characterize the kinds of functions learned by training with weight decay for multi-output (vector-valued) ReLU neural networks. This extends previous characterizations that were limited to single-output (scalar-valued) networks. This characterization requires the definition of a new class of neural function spaces that we call vector-valued variation (VV) spaces. We prove that neural networks (NNs) are optimal solutions to learning problems posed over VV spaces via a novel representer theorem. This new representer theorem shows that solutions to these learning problems exist as vector-valued neural networks with widths bounded in terms of the number of training data. Next, via a novel connection to the multi-task lasso problem, we derive new and tighter bounds on the widths of homogeneous layers in DNNs. The bounds are determined by the effective dimensions of the training data embeddings in/out of the layers. This result sheds new light on the architectural requirements for DNNs. Finally, the connection to the multi-task lasso problem suggests a new approach to compressing pre-trained networks.
We study the problem of parameter estimation in time series stemming from general stochastic processes, where the outcomes may exhibit arbitrary temporal correlations. In particular, we address the question of how much Fisher information is lost if the stochastic process is compressed into a single histogram, known as the empirical distribution. As we show, the answer is non-trivial due to the correlations between outcomes. We derive practical formulas for the resulting Fisher information for various scenarios, from generic stationary processes to discrete-time Markov chains to continuous-time classical master equations. The results are illustrated with several examples.
We study the problem of decentralized power allocation in a multi-access channel (MAC) with non-cooperative users, additive noise of arbitrary distribution and a generalized power constraint, i.e., the transmit power constraint is modeled by an upper bound on $\mathbb{E}[\phi(|S|)]$, where $S$ is the transmit signal and $\phi(.)$ is some non-negative, increasing and bounded function. The generalized power constraint captures the notion of power for different wireless signals such as RF, optical, acoustic, etc. We derive the optimal power allocation policy when there a large number of non-cooperative users in the MAC. Further, we show that, once the number of users in the MAC crosses a finite threshold, the proposed power allocation policy of all users is optimal and remains invariant irrespective of the actual number of users. We derive the above results under the condition that the entropy power of the MAC, $e^{2h(S)+c}$, is strictly convex, where $h(S)$ is the maximum achievable entropy of the transmit signal and $c$ is a finite constant corresponding to the entropy of the additive noise.
Kakutani's Fixed Point theorem is a fundamental theorem in topology with numerous applications in game theory and economics. Computational formulations of Kakutani exist only in special cases and are too restrictive to be useful in reductions. In this paper, we provide a general computational formulation of Kakutani's Fixed Point Theorem and we prove that it is PPAD-complete. As an application of our theorem we are able to characterize the computational complexity of the following fundamental problems: (1) Concave Games. Introduced by the celebrated works of Debreu and Rosen in the 1950s and 60s, concave $n$-person games have found many important applications in Economics and Game Theory. We characterize the computational complexity of finding an equilibrium in such games. We show that a general formulation of this problem belongs to PPAD, and that finding an equilibrium is PPAD-hard even for a rather restricted games of this kind: strongly-concave utilities that can be expressed as multivariate polynomials of a constant degree with axis aligned box constraints. (2) Walrasian Equilibrium. Using Kakutani's fixed point Arrow and Debreu we resolve an open problem related to Walras's theorem on the existence of price equilibria in general economies. There are many results about the PPAD-hardness of Walrasian equilibria, but the inclusion in PPAD is only known for piecewise linear utilities. We show that the problem with general convex utilities is in PPAD. Along the way we provide a Lipschitz continuous version of Berge's maximum theorem that may be of independent interest.
We introduce Markov Neural Processes (MNPs), a new class of Stochastic Processes (SPs) which are constructed by stacking sequences of neural parameterised Markov transition operators in function space. We prove that these Markov transition operators can preserve the exchangeability and consistency of SPs. Therefore, the proposed iterative construction adds substantial flexibility and expressivity to the original framework of Neural Processes (NPs) without compromising consistency or adding restrictions. Our experiments demonstrate clear advantages of MNPs over baseline models on a variety of tasks.
A randomized algorithm for a search problem is *pseudodeterministic* if it produces a fixed canonical solution to the search problem with high probability. In their seminal work on the topic, Gat and Goldwasser posed as their main open problem whether prime numbers can be pseudodeterministically constructed in polynomial time. We provide a positive solution to this question in the infinitely-often regime. In more detail, we give an *unconditional* polynomial-time randomized algorithm $B$ such that, for infinitely many values of $n$, $B(1^n)$ outputs a canonical $n$-bit prime $p_n$ with high probability. More generally, we prove that for every dense property $Q$ of strings that can be decided in polynomial time, there is an infinitely-often pseudodeterministic polynomial-time construction of strings satisfying $Q$. This improves upon a subexponential-time construction of Oliveira and Santhanam. Our construction uses several new ideas, including a novel bootstrapping technique for pseudodeterministic constructions, and a quantitative optimization of the uniform hardness-randomness framework of Chen and Tell, using a variant of the Shaltiel--Umans generator.
Game theory has by now found numerous applications in various fields, including economics, industry, jurisprudence, and artificial intelligence, where each player only cares about its own interest in a noncooperative or cooperative manner, but without obvious malice to other players. However, in many practical applications, such as poker, chess, evader pursuing, drug interdiction, coast guard, cyber-security, and national defense, players often have apparently adversarial stances, that is, selfish actions of each player inevitably or intentionally inflict loss or wreak havoc on other players. Along this line, this paper provides a systematic survey on three main game models widely employed in adversarial games, i.e., zero-sum normal-form and extensive-form games, Stackelberg (security) games, zero-sum differential games, from an array of perspectives, including basic knowledge of game models, (approximate) equilibrium concepts, problem classifications, research frontiers, (approximate) optimal strategy seeking techniques, prevailing algorithms, and practical applications. Finally, promising future research directions are also discussed for relevant adversarial games.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191