Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Architecture optimization techniques for Convolutional Neural Networks: further experiments and insights

Sobolewski Artur, Szyc Kamil

International Journal of Electronics and Telecommunications

2024

Vol. 70, No. 1

87--94

In this paper, we have researched implementing convolutional neural network (CNN) models for devices with limited resources, such as smartphones and embedded computers. To optimize the number of parameters of these models, we studied various popular methods that would allow them to operate more efficiently. Specifically, our research focused on the ResNet-101 and VGG-19 architectures, which we modified using techniques specific to model optimization. We aimed to determine which approach would work best for particular requirements for a maximum accepted accuracy drop. Our contribution lies in the comprehensive ablation study, which presents the impact of different approaches on the final results, specifically in terms of reducing model parameters, FLOPS, and the potential decline in accuracy. We explored the feasibility of implementing architecture compression methods that can influence the model’s structure. Additionally, we delved into post-training methods, such as pruning and quantization, at various model sparsity levels. This study builds upon our prior research to provide a more comprehensive understanding of the subject matter at hand.

Risk assessment of unfavorable interorganizational relationships in CSR projects considering the logistic aspects

Krechowicz Maria, Kiliańska Katarzyna, Sobolewski Artur

Zeszyty Naukowe. Organizacja i Zarządzanie / Politechnika Śląska

2022

z. 162

391--416

Purpose: The aim of this paper is to present a new model for risk assessment of unfavorable interorganizational relationships, among other things, in ventures classified as corporate social responsibility (CSR) projects. Design/methodology/approach: Scenario analysis, brainstorm sessions, literature study and own observations of interorganizational projects were used to develop a list of unwanted events and factors determining their occurrence. In the proposed risk assessment model, fault tree analysis and fuzzy logic were applied for qualitative and quantitative risk analysis. Thanks to applying the elements of fuzzy sets theory, it was possible to decrease the uncertainty and lack of precision in obtaining crisp values of the basic events’ probability. Findings: In this work 13 basic events and 41 risk factors determining occurrence of unfavorable interorganizational relationships in ventures were identified and described. The proposed model enabled to carry out qualitative and quantitative risk assessment of unfavorable interorganizational relationships projects. Its practical application was shown in an example of interorganizational CSR project concerning the organization of a mass event, considering its logistics aspects. Research limitations/implications: It is necessary to involve experts in risk assessment. This could be overcome by applying machine learning in future research. Practical implications: The application of the proposed model allows to effectively identify the critical risks, which should be of particular attention during the risk treatment stage. It aims to give a helping hand to all managers and practitioners who want to deliver attainable and successful interorganizational projects, supporting meeting the expectation of the engaged stakeholders. Social implications: Socially responsible activities contribute to solving and counteracting social problems. Originality/value: A novel risk assessment model of unfavorable interorganizational relationships in which 13 basic events and 41 risk determinants were considered. The model was presented at ventures classified as Corporate Social Responsibility projects.