Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 94 for “"Sum-Product"”.
-
The sum-product problem
The sum-product problem of Erdos and Szemeredi asserts that any subset of the integers has many products or many sums. We explore quantitative aspects of the problem over both the real numbers and finite fields of prime order.
-
An Empirical Study on Sum-Product Networks Structure Learning and Deep Convolutional Sum-Product Networks
Sum-product Networks (SPNs) are generative probabilistic models which obtained the attention of the community for its tractability and easy methods for learning. Our rst main contribution is an empirical comparison of methods for SPN learning and inference. LearnSPN is a popular algorithm for …
-
The Sum-Product Theorem and its Applications
… deep learning. I first identify and prove the sum-product theorem, which states that in any semiring for inference to be tractable it suffices that the factors of every product have disjoint scopes; i.e., that they are decomposable. I show that this simple condition unifies and extends many …
-
Parameter and Structure Learning Techniques for Sum Product Networks
… many variables. Among the various types of PGMs, sum-product networks (SPNs) have recently generated some interest because exact inference can always be done in linear time with respect to the size of the network. This is particularly attractive since it means that learning an SPN from data always …
-
On the Development of Deep Convolutional Sum-Product Networks
… join tree inference algorithm in BNs. A sum-product network (SPN) is another type of PGM defined by a DAG and a set of parameters. An SPN permits tractable inference, while inference is generally NP-hard in BNs. Furthermore, SPNs can be compiled from BNs or learned from data. In this …
-
Robustness in sum-product networks: from measurement to ensembles
… the problem of trust in classifications using Sum-Product Networks. A method of gauging the reliability of a classification through perturbing model weights using Credal Sum-Product Networks and creating a metric in the form of robustness to represent this is presented and demonstrated …
-
On the Relationship between Sum-Product Networks and Bayesian Networks
Sum-Product Networks (SPNs), which are probabilistic inference machines, have attracted a lot of interests in recent years. They have a wide range of applications, including but not limited to activity modeling, language modeling and speech modeling. Despite their practical applications and …
-
New Message Passing Methods for Min-Max and Sum-Product Inference
… problems. The second mode of interest is sum-product inference. Here, loopy BP performs poorly when the underlying mass contains multiple disjoint modes, capturing only one mode when it reaches a fixed point. As such, a message passing procedure that attempts to model all the fixed points …
-
Algorithms for Learning the Structure of Monotone and Nonmonotone Sum-Product Networks
The sum-product network (SPN) is a recently-proposed generative, probabilistic model that is guaranteed to compute any joint or any marginal probability in time linear in the size of the model. An SPN is represented as a directed, acyclic graph (DAG) of sum and product nodes, with univariate …
-
Probabilistic inference via sum-product algorithms on binary pairwise Gibbs random fields with applications to multiple fault diagnosis
… we study the application of iterative heuristic sum-product algorithms (SPAs) to the underlying graphs for solving the marginal problem on BPW-GRFs. These algorithms operate on the BPW-GRF graph by propagating messages along the edges and by using them to update the beliefs at each node of the …
-
Incidence Geometry in the Plane and Applications to Arithmetic Combinatorics
… in the plane and discuss its applications. The sum-product phenomenon is a motivating problem of this thesis and we discuss this in both the case of an arbitrary field and the specific case of F= R. In this latter setting, in Chapter 6 we prove a bound on the energy formulation of the …
-
Advances in Iterative Probabilistic Processing for Communication Receivers
… of the unknown channel or signal parameters. The sum-product algorithm is capable of efficiently approximating the marginal posterior probabilities desired for MAP detection and provides a unifying framework for the development of iterative receiver algorithms. However, in some applications the …
-
Geometric combinatorics in discrete settings
… sets in Fd/q and discuss some applications to sum-product problems. Finally, we study geometric combinatorics in Z/qZ. We generalize a result of Hart and Iosevich [27] which has applications to sum-product problems. Finally, we show that the Zd/q analogue of a sphere with unital radius is …
-
Directly Learning Tractable Models for Sequential Inference and DecisionMaking
… models such as arithmetic circuits (ACs) and sum-product networks (SPNs) for which inference is guaranteed to be tractable (e.g., linear in the size of the network for SPNs and ACs). This means that the networks learned from data can be directly used for inference without any further …
-
Learning Robust Tractable Models for Vision
… tackle these four desiderata, weaving together sum-product networks, neural networks, kernel machines, and symmetry group theory. First, we extend sum-product networks so that they can be trained discriminatively. This expands the space of SPN architectures and allows feature functions, making …
-
Multifold sums and products over R, and combinatorial problems on sumsets
We prove a new bound on a version of the sum-product problem studied by Chang. By introducing several combinatorial tools, this expands upon a method of Croot and Hart which used the Tarry-Escott problem to build distinct sums from polynomials with specific vanishing properties. We also study other …
-
Message passing algorithms - methods and applications
… which is often solved or approximated with Sum-Product belief propagation (BP) over the edges of the factor graph. For the case of sensor networks, where the conservation of energy is of critical importance and communication overhead can quickly drain this valuable resource, we present …
Page 1 of 5