Papers by Mircea Vladutiu
One of the main motivations behind social network analysis is the quest for understanding opinion... more One of the main motivations behind social network analysis is the quest for understanding opinion formation and diffusion. Previous models have limitations, as they typically assume opinion interaction mechanisms based on thresholds which are either fixed or evolve according to a random process that is external to the social agent. Indeed, our empirical analysis on large real-world datasets such as Twitter, Meme Tracker, and Yelp, uncovers previously unaccounted for dynamic phenomena at population-level, namely the existence of distinct opinion formation phases and social balancing. We also reveal that a phase transition from an erratic behavior to social balancing can be triggered by network topology and by the ratio of opinion sources. Consequently, in order to build a model that properly accounts for these phenomena, we propose a new (individual-level) opinion interaction model based on tolerance. As opposed to the existing opinion interaction models, the new tolerance model assumes that individual's inner willingness to accept new opinions evolves over time according to basic human traits. Finally, by employing discrete event simulation on diverse social network topologies, we validate our opinion interaction model and show that, although the network size and opinion source ratio are important, the phase transition to social balancing is mainly fostered by the democratic structure of the small-world topology.
Social Network Analysis and Mining, Oct 25, 2019
Simulating the behavior of economic agents fosters the analysis of interconnected markets dynamic... more Simulating the behavior of economic agents fosters the analysis of interconnected markets dynamics. Here, we extend the state of the art by adding realistic details to simulating economic exchange networks. To this end, we use our economic network simulation framework TrEcSim, which is designed to support the following real-life features: complex network topologies, evolution of economic agent roles, dynamic creation of new economic agents, diversity in product types, dynamic evolution of product prices, and investment decisions at agent level. By employing simulation, we determine which topological properties promote meritocracy and fairness. Simulation also allows for analyzing the influence of producers and middlemen distribution in the economic exchange network; similarly, we gain valuable insight regarding the distribution of payoff for each agent role. Moreover, we conclude that economic networks promote fairness throughout their structure, namely that the main determining factor for fairness in payoff distribution is the underlying network topology, not agent role assignment.
IEEE Conference Proceedings, 2018
This paper presents a design for identifying and classifying the Romanian traditional motifs foun... more This paper presents a design for identifying and classifying the Romanian traditional motifs found on 4 different categories (clothes, ceramics, carpets and painted eggs) by training a Convolutional Neural Network (CNN) model derived from the Residual Network (ResNet-50) architecture. We also implemented a system which can detect and identify through a webcam if the object in front of it contains a learned motif. Experimental results show that our neural network has an overall accuracy of 99.4% and a reduced webcam processing time.
Simulating the behavior of economic agents fosters the analysis of interconnected markets' dy... more Simulating the behavior of economic agents fosters the analysis of interconnected markets' dynamics. Here, we extend the state-of-the-art by adding realistic details to simulating economic exchange networks. To this end, we use our economic network simulation framework TrEcSim, which is designed to support the following real-life features: specific complex network topologies, evolution of economic agent roles, dynamic creation of new economic agents, diversity in product types, dynamic evolution of product prices, and investment decisions at agent-level. Using TrEcSim, we simulate and determine the point at which the networks (having different topology types) transition from being a topocratic system to becoming a meritocratic one. Simulation also allows for analyzing the dynamic evolution of producers and middlemen distribution in the economic exchange network. Moreover, we gain valuable insight regarding the distribution of payoff for each agent-role in various economic exchange networks, as follows: when producers are assigned randomly to topological positions, the payoff distribution within the producers category is fat-tailed (only a handful of producers benefit from an increased payoff), while the payoff of the middlemen category closely resembles a normal (Gaussian) distribution. However, when the topological positions of producers are assigned preferentially, the payoff distributions of the two role categories reverse.
This paper presents a method for identifying 34 animal classes corresponding to the most conventi... more This paper presents a method for identifying 34 animal classes corresponding to the most conventional animals found in the domestic areas of Europe by using four types of Convolutional Neural Networks (CNNs), namely VGG-19, InceptionV3, ResNet-50, and MobileNetV2. We also built a system capable of classifying all these 34 animal classes from images as well as in real-time from videos or a webcam. Additionally, our system is capable to automatically generate two new datasets, one dataset containing textual information (i.e. animal class name, date and time interval when the animal was present in the frame) and one dataset containing images of the animal classes present and identified in videos or in front of a webcam. Our experimental results show a high overall test accuracy for all 4 proposed architectures (90.56% for VGG-19 model, 93.41% for InceptionV3 model, 93.49 for ResNet-50 model and 94.54% for MobileNetV2 model), proving that such systems enable an unobtrusive method for gathering a rich collection of information about the vast numbers of animal classes that are being identified such as providing insights about what animal classes are present at a given date and time in a certain area and how they look, resulting in valuable datasets especially for researchers in the area of ecology
Springer eBooks, Sep 3, 2007
We have built a simulator named, CDLR SPEC 2000. This simulator is based on traces for systems wi... more We have built a simulator named, CDLR SPEC 2000. This simulator is based on traces for systems with cache memories. With CDLR SPEC 2000 program we can study, for a memory hierarcy, the next parameters: maping function, block size, writing strategies, replacement algoritm, size of cache memory, number of cache sets (for the set associative caches), number of words form
This paper presents a real-time hardware testing design based on a hybrid approach between Flying... more This paper presents a real-time hardware testing design based on a hybrid approach between Flying Probe-Inspired In-Circuit Testing (FPICT) and Joint Test Action Group (JTAG) debugging techniques. The FPICT is used for testing the physical parameter values of our dual-axis solar tracking equipment composed of $1 \\times$ Optocoupler, $1 \\times$ Arduino UNO board, and $2 \\times$ L298N motor drivers. Due to the Arduino UNO board’s inexistent JTAG testing facilities, we replaced it with an STM32 development board. In order to gain access to the internal logic of the STM32 microcontroller, we connected a dedicated and low-cost ST-Link V2 JTAG adapter to the FPICT device. Additionally, in order to make all 5 Circuits Under Tests (CUTs) test points (TPs) more accessible to the FPICT device, we designed a custom modular Printed Circuit Board (PCB). Finally, we present an autonomous design that achieves self-sufficiency regarding energy needs for the proposed hybrid testing method and the entire solar tracking equipment by making use entirely of solar energy. The experimental results show that the proposed non-intrusive hybrid testing approach is efficient regarding global fault coverage (66.35% for all targeted faults), placement accuracy (100% for considered test cases), testing time, and cost points of view.
An, offline fault detection strategy for the finite field inversion operation of Advanced Encrypt... more An, offline fault detection strategy for the finite field inversion operation of Advanced Encryption Standard (AES) is presented in this paper. The AES inversion units are stimulated using conventional Test Pattern Generation methods and the results' correctness is evaluated by means of signature analysis. The proposed architecture rely on connecting several inversion operation together in order to reduce the area overhead of the detection circuitry. The bytes of the AES state matrix are permuted sequentially in order to propagate the potential effect of a defect and reduce the number of signature registers required for detecting the presence of the defect. The proposed design is proved to represent an efficient test architecture by experimental results.
Lecture Notes in Computer Science, 2020
Climate change is considered to be one of the most important issues we are facing right now as a ... more Climate change is considered to be one of the most important issues we are facing right now as a specie and existent metrics and benchmarks used to evaluate the performance of different Deep Learning (DL) models and systems are currently focused mainly on their accuracy and speed, without also considering their energy consumption and cost. In this paper, we introduce four novel DL metrics, two regarding inference called Accuracy Per Consumption (APC) and Accuracy Per Energy Cost (APEC) and two regarding training called Time To Closest APC (TTCAPC) and Time To Closest APEC (TTCAPEC), which take into account not only a DL model's accuracy but also its energy consumption, energy cost and the time it takes to train it up to that point. Experimental results prove that all four DL metrics are promising, encouraging future DL researchers to make use of models and platforms that require low power consumption as well as of green energy when powering their DL-based systems.
2018 IEEE 24th International Symposium for Design and Technology in Electronic Packaging (SIITME), 2018
This paper presents a design for identifying and classifying the Romanian traditional motifs foun... more This paper presents a design for identifying and classifying the Romanian traditional motifs found on 4 different categories (clothes, ceramics, carpets and painted eggs) by training a Convolutional Neural Network (CNN) model derived from the Residual Network (ResNet-50) architecture. We also implemented a system which can detect and identify through a webcam if the object in front of it contains a learned motif. Experimental results show that our neural network has an overall accuracy of 99.4% and a reduced webcam processing time.
2014 18th International Conference on System Theory, Control and Computing (ICSTCC), 2014
We present a novel approach in designing and deploying traffic light systems by identifying key i... more We present a novel approach in designing and deploying traffic light systems by identifying key intersections of the road network. Based on techniques borrowed from Complex Network Analysis, our algorithm can be applied successively at different levels of granularity allowing a hierarchical clustering of the intersections and prioritization of the traffic lights. We illustrate our approach with a case study conducted over the city of Timisoara.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this p... more The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein.
A distribution-graph based scheduling algorithm is proposed together with an extended tree growin... more A distribution-graph based scheduling algorithm is proposed together with an extended tree growing technique to deal with the problem of unequal-length block-test scheduling under power dissipation constraints. The extended tree growing technique is used in combination with the classical scheduling approach in order to improve the test concurrency having assigned power dissipation limits. Its goal is to achieve a balanced test power dissipation by employing a least mean square error function. The least mean square error function is a distribution-graph based global priority function. Test scheduling examples and experiments highlight in the end the efficiency of this approach towards a system-level test scheduling algorithm
Declarations can be found on page 12 DOI 10.7717/peerj.722 Copyright 2015 Cecchin et al.
Twelfth International Conference on Machine Vision (ICMV 2019), 2020
This paper presents a method for detecting receipt fraud by implementing an Object Character Reco... more This paper presents a method for detecting receipt fraud by implementing an Object Character Recognition (OCR) algorithm composed of Image Processing Techniques and Convolutional Neural Networks (CNNs). We implemented two CNN models into a smartphone application that gives customers the option to take pictures of products they intend to buy (also to crop their price tags) while present in a hypermarket/supermarket as well as of the paid receipt and succeeds to automatically identify and compare all prices (multiple digits including decimals) of the products seen at the shelf and all prices found in the paid receipt, received from the cashier. This application helps the customer detect a receipt fraud due to a computer or human error, in a cheap and convenient way. Experimental results show 99.96% overall test accuracy for the CNN responsible for identifying product prices and 99.35% overall test accuracy for the CNN responsible for identifying receipt prices.
10th Euromicro Conference on Digital System Design Architectures, Methods and Tools (DSD 2007), 2007
This paper proposes a novel approach for increasing the performance of the floating point additio... more This paper proposes a novel approach for increasing the performance of the floating point addition, by efficiently exploiting both paths from the classical double path adder. Thus, it becomes possible to execute two floating point additions simultaneously using a single adder, each ...
… Conference on Artificial …, 2008
Abstract. This paper presents a new methodology together with its corresponding software analysis... more Abstract. This paper presents a new methodology together with its corresponding software analysis that create incentives for quantum cir-cuit synthesis. Our circuit synthesis and architecture design approaches bring a new view on quantum circuit synthesis. This paper presents a ...
2018 IEEE 24th International Symposium for Design and Technology in Electronic Packaging (SIITME), 2018
This paper presents a novel technique in testing the software code of a solar tracking device by ... more This paper presents a novel technique in testing the software code of a solar tracking device by implementing a White-box testing approach that makes use of a Wi-Fi module. First, we succeed in verifying if the wireless data transfer controlling the movements of the solar tracking device are in correspondence with the software code run on the main control board called Arduino UNO. Additionally to the local implementation, a cloud-based solution together with a mobile application to remotely control, test and communicate with the solar tracking device is proposed. Second, we used White-box testing techniques to test and give details about software errors in our solar tracker. We implemented unit testing techniques as well as custom code in order to find out all the loopholes and possible breakpoints in our solar tracker software by investigating Communication, Control Flow and Error handling errors. The experimental results show that our White-box testing strategy is efficient from t...
Uploads
Papers by Mircea Vladutiu