arXiv subjects

Balázs Csanád Csáji

Publications and source records attributed to Balázs Csanád Csáji.

25 records · Page 2Linked to original sources

arXiv2021

Exact Distribution-Free Hypothesis Tests for the Regression Function of Binary Classification via Conditional Kernel Mean Embeddings

In this paper we suggest two statistical hypothesis tests for the regression function of binary classification based on conditional kernel mean embeddings. The regression function is a fundamental object in classification as it determines both the Bayes optimal classifier and the misclassification probabilities. A resampling based framework is presented and combined with consistent point estimators of the conditional kernel mean map, in order to construct distribution-free hypothesis tests. These tests are introduced in a flexible manner allowing us to control the exact probability of type I error for any sample size. We also prove that both proposed techniques are consistent under weak statistical assumptions, i.e., the type II error probabilities pointwise converge to zero.

stat.ML↗

arXiv2019

Semi-Parametric Uncertainty Bounds for Binary Classification

The paper studies binary classification and aims at estimating the underlying regression function which is the conditional expectation of the class labels given the inputs. The regression function is the key component of the Bayes optimal classifier, moreover, besides providing optimal predictions, it can also assess the risk of misclassification. We aim at building non-asymptotic confidence regions for the regression function and suggest three kernel-based semi-parametric resampling methods. We prove that all of them guarantee regions with exact coverage probabilities and they are strongly consistent.

stat.ML↗

arXiv2018

Score Permutation Based Finite Sample Inference for Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) Models

A standard model of (conditional) heteroscedasticity, i.e., the phenomenon that the variance of a process changes over time, is the Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) model, which is especially important for economics and finance. GARCH models are typically estimated by the Quasi-Maximum Likelihood (QML) method, which works under mild statistical assumptions. Here, we suggest a finite sample approach, called ScoPe, to construct distribution-free confidence regions around the QML estimate, which have exact coverage probabilities, despite no additional assumptions about moments are made. ScoPe is inspired by the recently developed Sign-Perturbed Sums (SPS) method, which however cannot be applied in the GARCH case. ScoPe works by perturbing the score function using randomly permuted residuals. This produces alternative samples which lead to exact confidence regions. Experiments on simulated and stock market data are also presented, and ScoPe is compared with the asymptotic theory and bootstrap approaches.

stat.ME↗

arXiv2018

Wireless Multi-Sensor Networks for Smart Cities: A Prototype System with Statistical Data Analysis

As urbanization proceeds at an astonishing rate, cities have to continuously improve their solutions that affect the safety, health and overall wellbeing of their residents. Smart city projects worldwide build on advanced sensor, information and communication technologies to help dealing with issues like air pollution, waste management, traffic optimization, and energy efficiency. The paper reports about the prototype of a smart city initiative in Budapest which applies various sensors installed on the public lighting system and a cloud-based analytical module. While the installed wireless multi-sensor network gathers information about a number of stressors, the module integrates and statistically processes the data. The module can handle inconsistent, missing and noisy data and can extrapolate the measurements in time and space, namely, it can create short-term forecasts and smoothed maps, both accompanied by reliability estimates. The resulting database uses geometric representations and can serve as an information centre for public services.

cs.CY↗

arXiv2015

Cooperative Control in Production and Logistics

Classical applications of control engineering and information and communication technology (ICT) in production and logistics are often done in a rigid, centralized and hierarchical way. These inflexible approaches are typically not able to cope with the complexities of the manufacturing environment, such as the instabilities, uncertainties and abrupt changes caused by internal and external disturbances, or a large number and variety of interacting, interdependent elements. A paradigm shift, e.g., novel organizing principles and methods, is needed for supporting the interoperability of dynamic alliances of agile and networked systems. Several solution proposals argue that the future of manufacturing and logistics lies in network-like, dynamic, open and reconfigurable systems of cooperative autonomous entities. The paper overviews various distributed approaches and technologies of control engineering and ICT that can support the realization of cooperative structures from the resource level to the level of networked enterprises. Standard results as well as recent advances from control theory, through cooperative game theory, distributed machine learning to holonic systems, cooperative enterprise modelling, system integration, and autonomous logistics processes are surveyed. A special emphasis is put on the theoretical developments and industrial applications of Robustly Feasible Model Predictive Control (RFMPC). Two case studies are also discussed: i) a holonic, PROSA-based approach to generate short-term forecasts for an additive manufacturing system by means of a delegate multi-agent system (D-MAS); and ii) an application of distributed RFMPC to a drinking water distribution system.

eess.SY↗

arXiv2014

Adaptive Stochastic Resource Control: A Machine Learning Approach

The paper investigates stochastic resource allocation problems with scarce, reusable resources and non-preemtive, time-dependent, interconnected tasks. This approach is a natural generalization of several standard resource management problems, such as scheduling and transportation problems. First, reactive solutions are considered and defined as control policies of suitably reformulated Markov decision processes (MDPs). We argue that this reformulation has several favorable properties, such as it has finite state and action spaces, it is aperiodic, hence all policies are proper and the space of control policies can be safely restricted. Next, approximate dynamic programming (ADP) methods, such as fitted Q-learning, are suggested for computing an efficient control policy. In order to compactly maintain the cost-to-go function, two representations are studied: hash tables and support vector regression (SVR), particularly, nu-SVRs. Several additional improvements, such as the application of limited-lookahead rollout algorithms in the initial phases, action space decomposition, task clustering and distributed sampling are investigated, too. Finally, experimental results on both benchmark and industry-related data are presented.

cs.LG↗

arXiv2012

PageRank Optimization by Edge Selection

The importance of a node in a directed graph can be measured by its PageRank. The PageRank of a node is used in a number of application contexts - including ranking websites - and can be interpreted as the average portion of time spent at the node by an infinite random walk. We consider the problem of maximizing the PageRank of a node by selecting some of the edges from a set of edges that are under our control. By applying results from Markov decision theory, we show that an optimal solution to this problem can be found in polynomial time. Our core solution results in a linear programming formulation, but we also provide an alternative greedy algorithm, a variant of policy iteration, which runs in polynomial time, as well. Finally, we show that, under the slight modification for which we are given mutually exclusive pairs of edges, the problem of PageRank optimization becomes NP-hard.

cs.DS↗