Jarosław Rudy

A new approach for multi-criteria scheduling

Display Omitted We use a Enhanced SA algorithm for bi-criteria permutation flow shop problem.We consider vector processing to improve computational time of the algorithm.Implementation of VESA is achieved even on single thread processors.Considerable increase in number of evaluated solutions per unit of time is observed.Better approximation of the Pareto frontier is obtained. In this paper we propose a new approach, recommended for solving certain class of multi-criteria scheduling problems, with the use of cloud exploration of the solution space supported by fine-grained parallel computing. The approach allow us to approximate Pareto front more accurately than other known algorithms in competitive time. To show and check advantageous properties of the proposed approach, the new solution algorithm, called VESA, was implemented for the case of bi-criteria flow shop scheduling problem and tested against a number of high-quality benchmarks known in the literature. Vector processing technologies are used to enhance the efficiency of solution search and acceptance rates for the extended simulated annealing metaheuristic. Results are compared using, among others, the independent Hyper-Volume Indicator ( I H ) measure. Computer test of VESA confirms excellent approximation of the Pareto front.

Cyclic Scheduling Line with Uncertain Data

In this paper, a cyclic permutation flow shop problem for a certain production line with uncertain data is considered. The goal is to minimize the cycle time. The uncertain elements in the system are identified and modeled as fuzzy numbers. A metaheuristic fuzzy-aware algorithm is developed and tested against 3 deterministic algorithms. The fuzzy algorithm significantly outperforms deterministic algorithms 70 % of the time with similar computation time. The fuzzy algorithm is also more reliable, providing solutions with smaller standard deviation.

GACO: A Parallel Evolutionary Approach to Multi-objective Scheduling

In this paper the job shop scheduling problem with two criteria of minimizing makespan and the sum of tardiness of jobs is considered. This multi-objective problem is strongly NP-hard, as single criterion version is strongly NP-hard as well. A permutation-based representation for the job shop problem is used and a new hybrid parallel multi-agent method, called GACO (Genetic Algorithm Ant Colony Optimization), is proposed. The computation is done in parallel and additional threads concurrently compute certain parts of both algorithms. The researched speed-up is considerable, albeit limited by the need to combine solutions. Approximation of the Pareto front obtained by GACO is superior to the approximations obtained by GA and ACO separately.

Dangerous events related to the use of scaffolding

Abstract The majority of construction works are carried out at height; these are works that carry a high risk of accident. This paper presents an analysis of dangerous events associated with scaffolding. Information about these events was obtained from online resources. The collected data not only allowed analysis of occupational accidents but also accidents involving third parties and near-miss events. The paper presents particular examples of these events.

Turing machine approach to runtime software adaptation

In this paper, the problem of applying changes to software at runtime is con- sidered. The computability theory is used in order to develop a more general and programming-language-independent model of computation with support for runtime changes. Various types of runtime changes were defined in terms of computable functions and Turing machines. The properties of such functions and machines were used to prove that arbitrary runtime changes on Turing machines are impossible in general cases. A method of Turing machine decom- position into subtasks was presented and runtime changes were defined through transformations of the subtask graph. Requirements for the possible changes were considered with regard to the possibility of subtask execution during such changes. Finally, a runtime change model of computation was defined by exten- sion of the Universal Turing Machine.

Dynamic Random-Access Stored-Program Machine for Runtime Code Modification

This paper is concerned with the study of possibility of performing changes to existing running programs with the use of the RAM and RASP models of computation. A new model of computation is defined with the capability of performing runtime changes. Theoretical properties, including time and space complexities, of the defined models are presented and proven. A number of simple empirical tests are conducted in order to prove the ability to perform runtime changes as well as support obtained theoretical results. The paper concludes that the defined model has virtually no affect on performance when there are no changes and the performance with changes is easily manageable. Moreover, the results can be used to develop runtime change capabilities for a wide range of programming languages and paradigms.

Models and Scheduling Algorithms for a Software Testing System Over Cloud

In this paper a Testing-as-a-Service cloud system is considered. The problem of distributing software tests over cloud is modeled as a variation of the online scheduling problem with the goal of minimizing the average flowtime of jobs. History of over 20 million tests is used to develop statistical and empirical models of system activity. Developed models are used to generate a number of problem instances. The instances are then used to test the effectiveness of several scheduling algorithms. Results indicate that SJLOF algorithm is superior to other tested algorithms, especially when the cloud resources are limited.

Lifelogging System Based on Averaged Hidden Markov Models: Dangerous Activities Recognition for Caregiver Support

In this paper a prototype lifelogging system for monitoring persons with cognitive disabilities and elderly people, as well as a method for automatic detection of dangerous activities are presented. The system allows remote monitoring of observed persons via Internet website and respects the privacy of the persons by displaying their silhouettes instead of actual images. Application allows viewing of both real-time and historic data. Lifelogging data (skeleton coordinates) needed for posture and activity recognition are acquired using Microsoft Kinect 2.0. Several activities are marked as potentially dangerous and generate alarms sent to the caregivers upon detection. Recognition models are developed using Averaged Hidden Markov Models with multiple learning sequences. Action recognition includes methods for differentiation between normal and potentially dangerous activities e.g. self-aggressive autistic behavior) using the same motion trajectory. Some activity recognition examples and results are presented.

Reversed Amdahl’s Law for Hybrid Parallel Computing

The paper presents a certain version of Amdahl’s law intended for use with heterogeneous systems of parallel computation, which can be used to compare different technologies and configurations of parallel systems in a better way than a simple comparison of achieved speedups. A mathematical formulation of the law, derived from the known Amdahl’s law is presented. Next, the proposed law is used in simple comparison of performance of several computation devices. Finally, we use the law to estimate the performance limit of a certain heterogeneous parallel system equipped with GPU device and employing vector processing techniques.

Dynamic Turing Machine: model and properties for runtime code changes

In this paper, a dynamic model of computation based on the Universal Turing Machine is proposed. This model is capable of applying runtime code modifications for 3-symbol deterministic Turing Machines at runtime and requires a decomposition of the simulated machine into parts called subtasks. The algorithm for performing runtime changes is considered, and the ability to apply runtime changes is studied through computer simulations. Theoretical properties of the proposed model, including computational power as well as time and space complexity, are studied and proven. Connections between the proposed model and Oracle Machines are discussed. Moreover, a possible method of implementation in real-life systems is proposed.