[신간의 별자리x] 우리/미술, 그리고 ‘슬픔의 박물관’

We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Random graphs

物件導向軟體之架構(Object-Oriented Software Construction)探討

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

With the continuous increase in complexity and expense of industrial systems, there is less tolerance for performance degradation, productivity decrease, and safety hazards, which greatly necessitates to detect and identify any kinds of potential abnormalities and faults as early as possible and implement real-time fault-tolerant operation for minimizing performance degradation and avoiding dangerous situations. During the last four decades, fruitful results have been reported about fault diagnosis and fault-tolerant control methods and their applications in a variety of engineering systems. The three-part survey paper aims to give a comprehensive review of real-time fault diagnosis and fault-tolerant control, with particular attention on the results reported in the last decade. In this paper, fault diagnosis approaches and their applications are comprehensively reviewed from model- and signal-based perspectives, respectively.

/pdf/a-survey-of-fault-diagnosis-and-fault-tolerant-techniques-9uuf1uqysf.pdf

A Survey of Fault Diagnosis and Fault-Tolerant Techniques—Part I: Fault Diagnosis With Model-Based and Signal-Based Approaches

Modern electric grids that integrate smart grid technologies require different approaches to grid operations. There has been a shift towards increased reliance on distributed sensors to monitor bidirectional power flows and machine learning based load forecasting methods (e.g., using deep learning). These methods are fairly accurate under normal circumstances, but become highly vulnerable to stealthy adversarial attacks that could be deployed on the load forecasters. This paper provides a novel model-based Testbed for Simulation-based Evaluation of Resilience (TeSER) that enables evaluating deep learning based load forecasters against stealthy adversarial attacks. The testbed leverages three existing technologies, viz. DeepForge: for designing neural networks and machine learning pipelines, GridLAB-D: for electric grid distribution system simulation, and WebGME: for creating web-based collaborative metamodeling environments. The testbed architecture is described, and a case study to demonstrate its capabilities for evaluating load forecasters is provided.

Online Testbed for Evaluating Vulnerability of Deep Learning Based Power Grid Load Forecasters

This paper presents the Distributed End-to-end Utiization CONtrol (DEUCON) algorithm. DEUCON can dynamically enforce desired CPU utilizations on all processors in a distributed real-time system despite uncertainties in the system workload. In contrast to earlier centralized control schemes, DEUCON is a distributed control algorithm that is systematically designed based on the Distributed Model Predictive Control theory. We decompose the global multi-processor utilization control problem into a set of localized subproblems, and design a peer-to-peer control structure where each local controller only needs to coordinate with a small number of neighbor processors. DEUCON can provide utilization guarantees similar to a centralized control algorithm, while significantly reducing the per-controller run-time overhead in terms of both computation and communication. Furthermore, it can tolerate considerable network delay and individual processor failures. Consequently, DEUCON can provide scalable and robust utilization control services for large distributed real-time systems that operate in unpredictable environments.

/pdf/deucon-distributed-end-to-end-utilization-control-for-real-31x84hpr3c.pdf

DEUCON: Distributed End-to-End Utilization Control for Real-Time Systems

This paper presents a novel framework for hierarchical control of piecewise linear hybrid dynamical systems. The main characteristic of this class of hybrid systems is that the continuous dynamics are described by linear difference equations, the discrete dynamics by finite automata, and the interaction between the continuous and the discrete part is defined by piecewise linear maps. Control design is formulated as a regulator problem and algorithms for the synthesis of dynamical controllers are developed. Control specifications are modeled as finite automata. Both static specifications that do not change as time progresses and dynamic specifications that include sequencing of events and eventual execution of actions are considered. Control design is implemented using finite automata and linear programming techniques. Simulation results of a tank system are used to illustrate the approach.

/pdf/design-of-piecewise-linear-hybrid-dynamical-systems-using-a-24r6b2gtex.pdf

Design of piecewise linear hybrid dynamical systems using a control regulator approach

Autonomous systems use extensively learning-enabled components such as deep neural networks (DNNs) for prediction and decision making. In this paper, we utilize a feedback loop between learning-enabled components used for classification and the sensors of an autonomous system in order to improve the confidence of the predictions. We design a classifier using Inductive Conformal Prediction (ICP) based on a triplet network architecture in order to learn representations that can be used to quantify the similarity between test and training examples. The method allows computing confident set predictions with an error rate predefined using a selected significance level. A feedback loop that queries the sensors for a new input is used to further refine the predictions and increase the classification accuracy. The method is computationally efficient, scalable to high-dimensional inputs, and can be executed in a feedback loop with the system in real-time. The approach is evaluated using a traffic sign recognition dataset and the results show that the error rate is reduced.

Improving Prediction Confidence in Learning-Enabled Autonomous Systems

This book attempts to bring together recent developments in hybrid systems research for formal analysis methods and software engineering tools for building large information systems. The result is the construction of a elaborate modeling formalism that is very expressive and flexible in order to allow modeling of real-world examples. The value of this book is counterbalanced by weaknesses with respect to theoretical foundations and methods, automated analysis tools. and the pedagogical value.

Xenofon Koutsoukos

Papers

Online Testbed for Evaluating Vulnerability of Deep Learning Based Power Grid Load Forecasters

DEUCON: Distributed End-to-End Utilization Control for Real-Time Systems

Design of piecewise linear hybrid dynamical systems using a control regulator approach

Improving Prediction Confidence in Learning-Enabled Autonomous Systems

Modeling and Analysis of Hybrid Supervisory Systems: A Petri Net Approach (by Villani, E. et al.; 2007) [Book review]