Prof. Sheng-Lung Peng
National Taipei University of Business
Title: Domination-like Problems with Propagation Property
Abstract
Influence maximization is an important problem in the fields of social networks and data mining. Propagation is one of the important properties of this problem. In graph theory, the power domination problem is one of the few problems with propagation properties. This study combines the concepts of influence maximization and power domination problems. We propose some problems with propagation properties. For example, in the k-influence optimization problem, our goal is to find a seed set with the smallest size such that they can spread and influence everyone on the graph through their influence. In the problem, a person is influenced if his/her k friends are influenced. In this research, we consider this propagation property on domination-like problems.
Prof. Chuan-Ju Wang
Academia Sinica
Title: Markov Chain Importance Sampling for Minibatches
Abstract
This study investigates importance sampling under the scheme of minibatch stochastic gradient descent, under which the contributions are twofold. First, theoretically, we develop a neat tilting formula, which can be regarded as a general device for asymptotically optimal importance sampling. Second, practically, guided by the formula, we present an effective algorithm for importance sampling which accounts for the effects of minibatches and leverages the Markovian property of the gradients between iterations. Experiments conducted on artificial data confirm that our algorithm consistently delivers superior performance in terms of variance reduction. Furthermore, experiments carried out on real-world data demonstrate that our method, when paired with relatively straightforward models like multilayer perceptron and convolutional neural networks, outperforms in terms of training loss and testing error.