327,213 research outputs found
Virtual private network design over the first Chvátal closure
No full textIn this paper we consider the virtual private network (VPN) design problem. Given upper bounds on the amount of traf- fic that an endpoint could send or receive, the problem needs to reserve enough capacities in such a way that any demand matrix that respects the upper bound could be routed without exceeding the reserved ca- pacities and the total reservation cost is minimized. In [A. Moradi, A. Lodi and S.M. Hashemi. Networks 63 (2014) 327–333.] it is argued that the computational investigation on exact mathematical program- ming approaches for VPN needs to be revised after that challenging instances have been exposed. To that end, we consider the VPN design problem over the first Chva ́tal closure and demonstrate that tight so- lutions could be found for the VPN design problem only by optimizing over the closure. First, we perform theoretical investigation on adding rank-1 Chva ́tal–Gomory cuts to the problem. Along the way, an im- portant property for such cuts is proved that omits a large number of redundant rank-1 cuts. We then provide interesting insights about the problem and reduce the existing MIP formulations to a binary one. On the computational side, we investigate the idea of adding rank-1 cuts more aggressively in order to computationally evaluate tightness of the first Chva ́tal closure for the VPN design problem. Here, the binary re- duction plays an important role allowing the use of special cuts of the first closure, namely the zero-half cuts. We show that, almost all the success of the first Chv ́atal closure of the VPN design problem in rais- ing dual bound is due to zero-half cuts. Our experiments on the bench- mark instances in [A. Moradi, A. Lodi and S.M. Hashemi. Networks 63 (2014) 327–333.] show that a state-of-the-art IP solver without using zero-half cuts could not even hit the challenging benchmarks. As a re- sults a cut-and-branch framework that aggressively adds such cuts at the root could solve the challenging VPN instances to the extent of zero or small integrality gap in a reasonable amount of time
A strain energy criterion based on grain dislodgment at borehole wall in poorly cemented sands
No full textAbstract not availableS. S. Hashemi, N. Melkoumia
Grey literature as Valuable Resources in National Library of Iran: From Organizing to Digitization
No full textGrey literature is an important source of information due to the uniqueness of the content that gets published. Because commercial publishers are looking to make a profit on the materials they publish, they often overlook niche research areas that serve smaller populations. Grey literature is one way to search for information in emerging or less popular research areas. It seems that it reinforce research and accelerate continuous movement towards having a scientific society. Any source of information that has not been released to the market after printing is considered as grey literature. Grey literature is a special resource that is created for a specific purpose and audience and is replicated finitely. It is impermanent and invisible, and cannot be seen in the directory of publishers, bookshops and libraries.
The National Library of Iran was founded in 1937. The main goal of this organization is collecting, preserving, organizing and disseminating information about printed and non-printed works in Iran, and taking measures and making decisions to guarantee the accuracy, ease and speed of research and study in all fields to promote national culture. In order to achieve these goals, and according to the law, all private and public publishers are required to submit a copy of their publications (book and non -book materials) to the National Library of Iran . There are more than 900 thousand issues in the grey literature group at the National Library, over 240,000 issues of which are digitized. After being collected, these resources are organized and made available to users. Report, research project, standard and dissertations are among the most important sources of the National Library of Iran ’s grey literature, used by many researchers daily.In this research, the process of collecting, organizing and disseminating of information of these sources in the National Library of Iran will be expressed as a successful practical experience. Besides, it will be shown how the National Library of Iran has dealt with the problems it faced with
Nothotylenchus siddiqii Hashemi & Karegar 2020, n. sp.
No full textNothotylenchus siddiqii n. sp. (Figures 3 & 4) Measurements: Table 2. Female: Body straight or slightly curved ventrally. Cuticular annulation distinct, about 1 µm wide; lateral fields with six to nine incisures, 5–7 µm wide and occupying 31.2–42.1% of body width. Head low, with rounded margins, not offset, with two or three fine annuli, 1.5–2 µm high and about 5.5 µm wide. Cephalic skeleton moderately developed, outer margin of basal plate extending two to three annuli inside body. Stylet delicate, conus 26.3–39.7% of total stylet length, knobs medium-sized or small and rounded, 1–2 µm wide. DGO 0.5–1 µm posterior to stylet knobs. Median pharyngeal bulb elongated and spindle shape, wider than procorpus, 4.5–6 µm wide, valveless. Isthmus cylindrical and more slender than procorpus. Nerve ring surrounding isthmus almost at its middle, 69 (65.5–71.5) µm from anterior end. Hemizonid about two to three cuticular annuli long, 82 (79.5–88.5) µm from anterior end, two annuli anterior to S-E pore. S-E pore located from posterior half of isthmus to junction of isthmus and basal bulb. Deirids at the level of S-E pore. Basal pharyngeal bulb pyriform to slightly elongate and offset from intestine, length to width ratio 2.3 (1.8–2.8), with the two anterior-most intestinal cells appearing hyaline. Oocytes in single row, spermatheca elongate, filled with round sperms or empty, uterus quadricolumellar, sometimes with a valve between spermatheca and uterus, vagina 4–6 µm in length, 27.0–34.9% of VBW. PUS well developed, 26.5–40 µm, 2.2 (1.9–2.4) VBW, 54.2 (46.3–61.3) % V-A and 3.1 (2.6–3.6) ABW. Post-vulval body length 10.6 (10.5–10.9) ABW. Tail conical with usually rounded or seldom dull terminus. Male: General morphology, stylet, pharynx and tail shape similar to females. Spicules ventrally arcuate, gubernaculum simple and crescent-shaped, bursa 12–23.5 µm long and enveloping 25.7–48.6% of tail length. Tail conical with rounded terminus. Habitat & locality. Type population collected in 2014 from rhizosphere of alfalfa in Darab region, Fars province, Iran (GPS coordinates: 28°41.167′N, 54°16.056′E, elevation 1367 m.). Type material. Female holotype, three female paratypes and six male paratypes on six glass slides kept in the nematode collection of the Department of Plant Protection, School of Agriculture, Shiraz University, Iran, and two female paratypes and three male paratypes on two glass slides deposited in the Nematode Collection of the Plantenziektenkundige Dienst, Wageningen, The Netherlands. Diagnosis and relationships. Nothotylenchus siddiqi n. sp. is characterised by its short body, 573–645 µm in length, six to nine lateral field incisures, short, delicate stylet (6.5–7.5 µm) with rounded knobs, pyriform or slightly elongate and offset basal pharyngeal bulb, posterior vulva position (V = 79.3–81.0), PUS = 26.5–40 µm, short spicules, 14.5–16.5 µm long, and tail with rounded terminus. N. siddiqii n. sp. resembles N. affinis, N. hexaglyphus, N. persicus, N taylori Husain & Khan, 1974; N. geraerti and N. medians. But it can be distinguished from all these species by having 6–9 incisures in lateral fields (vs. 6). Moreover N. siddiqii n. sp. differs from N. affinis by greater PUS/VBW ratio (1.9–2.4 vs. 1.1–1.3), from N. hexaglyphus by shorter spicules (14.5–16.5 vs. 19–22 μm) and greater PUS/VBW ratio (1.9–2.4 vs. 0.5–1.3), from N. taylori by relatively shorter spicules (14.5–16.5 vs. 16.5–20 μm) and from N. persicus by shorter spicules (14.5–16.5 vs. 21–22 μm), position of the S-E pore (posterior half of isthmus to its junction with basal pharyngeal bulb vs. after basal pharyngeal bulb) and the tail terminus shape (rounded vs. pointed). Etymology. The species is named in honor of Dr. Mohammad Rafiq Siddiqi for his outstanding contributions to the science of nematology.Published as part of Hashemi, Kobra & Karegar, Akbar, 2020, New and known species of Nothotylenchus Thorne, 1941 (Nematoda: Anguinidae) from Iran with an updated list of species, pp. 482-500 in Zootaxa 4729 (4) on pages 487-489, DOI: 10.11646/zootaxa.4729.4.2, http://zenodo.org/record/375201
Effect of different stress path regimes on borehole instability in poorly cemented granular formations
No full textAbstract not availableS.S. Hashemi, N. Melkoumia
Extraction of the Essential Constituents of S&P500 Index
No full textGeorgia Southern University faculty members Ray R. Hashemi, Omid Ardakani, Azita Bahrami, and Jeffery A. Young authored Extraction of the Essential Constituents of S&P500 Index in Fourth International Conference on Computational Science and Computational Intelligence (CSCI’17)
A borehole stability study by newly designed laboratory tests on thick-walled hollow cylinders
No full textAbstract not availableS.S. Hashemi, N. Melkoumian, A. Taher
Data-driven software system log anomaly detection
No full textAbstract
Software logs are semi-structured text files that preserve states and events from software runtime. They provide detailed records that help to diagnose issues, monitor performance, and ensure security. Due to the increasing complexity and volume of logs, manual inspection has become impractical, making automated log analysis essential for tasks such as anomaly detection, troubleshooting, and performance optimization. This thesis compiles four papers contributing to software log anomaly detection field. While the first three papers contribute to the four stage (Pre-processor, Parser, Vectorizer, and Classifier) pipeline, Paper IIII proposes a new alternative to it.
At first, the thesis introduces Paper I, an advanced log parser that fits into the parsing stage within the four-stage log anomaly detection system. The paper utilizes a rule-based tokenizer, interdependency token graph, and parallel processing to offer accurate and scalable log parsing. Moreover, the thesis dives deeper into log parsing with Paper II, introducing a new metric to measure character-level accuracy in log parsing alongside a new benchmark dataset tailored to that metric. This benchmark provides a more precise method of assessing and comparing log parsers' accuracy and performance.
For the last stage, the thesis introduces Paper III, which proposes a method built on the foundation of the Siamese networks. The paper differentiates itself from the competition by not only offering enhanced robustness against log evolution but also providing a solution to detect over-the-time data drifts in logs.
Additionally, Paper IIII replaces the parser, vectorizer, and classifier with a character-based Hierarchical Convolutional Neural Network (HCNN), transforming the multi-staged system into an end-to-end one. HCNN allows for analyzing the usually ignored digits, numbers, and punctuations alongside the natural language. Furthermore, Paper4 offers improved generalization across datasets, making it adaptable even in scenarios with limited training data.
Overall, the thesis contributes to software log anomaly detection in four papers. These contributions improve the system's accuracy, performance, scalability, and robustness across a diverse set of evaluations using public open datasets. Original papers Hashemi, S., & Mäntylä, M. (2024). Token interdependency parsing (Tipping)—Fast and accurate log parsing. Advance online publication. https://doi.org/10.48550/arXiv.2408.00645 https://doi.org/10.48550/arXiv.2408.00645 Hashemi, S., Nyyssölä, J., & Mäntylä, M. V. (2024). LogPM: Character-based log parser benchmark. 2024 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER), 705–710. https://doi.org/10.1109/SANER60148.2024.00077 https://doi.org/10.1109/SANER60148.2024.00077 Self-archived version Hashemi, S., & Mäntylä, M. (2022). SiaLog: Detecting anomalies in software execution logs using the siamese network. Automated Software Engineering, 29(2), 61. https://doi.org/10.1007/s10515-022-00365-7 https://doi.org/10.1007/s10515-022-00365-7 Self-archived version Hashemi, S., & Mäntylä, M. (2024). OneLog: Towards end-to-end software log anomaly detection. Automated Software Engineering, 31(2), 37. https://doi.org/10.1007/s10515-024-00428-x https://doi.org/10.1007/s10515-024-00428-x Self-archived version Tiivistelmä
Ohjelmistolokit ovat puolistrukturoituja tekstitiedostoja, jotka sisältävät ohjelmiston suorituksen tapahtumat. Ne tarjoavat yksityiskohtaisia tietoja, jotka auttavat vikojen diagnosoinnissa, suorituskyvyn seurannassa ja tietoturvan varmistamisessa. Lokien monimutkaisuus ja määrä tekevät manuaalisesta lokien tarkastelusta epäkäytännöllistä, minkä vuoksi automaattinen lokianalyysi on välttämätöntä. Tämä väitöskirja hyödyntää useista eri lähteistä kerättyjä julkisia lokiaineistoja, jotka kattavat laajasti erilaisia ohjelmistojärjestelmiä ja supertietokoneita. Väitöskirja tarjoaa ratkaisuja ohjelmistolokien poikkeuksien tunnistukseen.
Opinnäyte esittelee Artikkelissa I kehittyneen lokijäsentimen, joka sijoittuu nelivaiheisen lokianomalian tunnistusjärjestelmän jäsentämisvaiheeseen. Artikkeli hyödyntää sääntöpohjaista tokenointia, tokenien välisiä riippuvuuksia mallintavaa graafia ja rinnakkaislaskentaa tarjotakseen tarkan ja skaalautuvan lokijäsentimen.
Seuraavaksi opinnäyte syventyy lokijäsentämiseen Artikkelissa II, joka esittelee uuden mittarin lokijäsentämisen tarkkuuden arviointiin merkkitasolla sekä uuden vertailuaineiston. Tämä vertailuaineisto tarjoaa tarkemman tavan arvioida ja vertailla lokijäsentimien tarkkuutta ja suorituskykyä.
Artikkeli III ehdottaa menetelmää poikkeusten luokitteluun, joka pohjautuu Siamese-neuroverkkoihin. Tämä lähestymistapa eroaa aiemmista tutkimuksista tarjoamalla paitsi parempaa sietokykyä lokien muutoksia kohtaan ja myös keinon havaita pitkän aikavälin datasiirtymiä lokeissa.
Lisäksi opinnäyte esittelee uudenlaisen lähestymistavan Artikkelissa IV. Tässä ratkaisussa jäsennin, vektorisointivaihe ja luokittelija korvataan merkkipohjaisella hierarkkisella konvoluutioneuroverkolla (HCNN), mikä muuntaa monivaiheisen järjestelmän päästä-päähän -ratkaisuksi. HCNN mahdollistaa usein huomiotta jäävien numeroiden ja välimerkkien analyysin yhdessä luonnollisen kielen kanssa. Artikkeli IV parantaa myös yleistettävyyttä eri aineistojen välillä, tehden siitä sovellettavan myös tilanteissa, joissa opetusaineistoa on niukasti. Kaiken kaikkiaan opinnäyte esittelee neljä menetelmää ohjelmistolokien analysointiin. Menetelmät parantavat poikkeuksien tunnistamisen tarkkuutta, suorituskykyä, skaalautuvuutta ja vikasietoisuutta laajassa joukossa julkisia lokiaineistoja. Osajulkaisut Hashemi, S., & Mäntylä, M. (2024). Token interdependency parsing (Tipping)—Fast and accurate log parsing. Advance online publication. https://doi.org/10.48550/arXiv.2408.00645 https://doi.org/10.48550/arXiv.2408.00645 Hashemi, S., Nyyssölä, J., & Mäntylä, M. V. (2024). LogPM: Character-based log parser benchmark. 2024 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER), 705–710. https://doi.org/10.1109/SANER60148.2024.00077 https://doi.org/10.1109/SANER60148.2024.00077 Rinnakkaistallennettu versio Hashemi, S., & Mäntylä, M. (2022). SiaLog: Detecting anomalies in software execution logs using the siamese network. Automated Software Engineering, 29(2), 61. https://doi.org/10.1007/s10515-022-00365-7 https://doi.org/10.1007/s10515-022-00365-7 Rinnakkaistallennettu versio Hashemi, S., & Mäntylä, M. (2024). OneLog: Towards end-to-end software log anomaly detection. Automated Software Engineering, 31(2), 37. https://doi.org/10.1007/s10515-024-00428-x https://doi.org/10.1007/s10515-024-00428-x Rinnakkaistallennettu versio Academic dissertation to be presented with the assent of the Doctoral Programme Committee of Information Technology and Electrical Engineering of the University of Oulu for public defence in the OP auditorium (L10), Linnanmaa, on 12 June 2025, at 12 noonAbstract
Software logs are semi-structured text files that preserve states and events from software runtime. They provide detailed records that help to diagnose issues, monitor performance, and ensure security. Due to the increasing complexity and volume of logs, manual inspection has become impractical, making automated log analysis essential for tasks such as anomaly detection, troubleshooting, and performance optimization. This thesis compiles four papers contributing to software log anomaly detection field. While the first three papers contribute to the four stage (Pre-processor, Parser, Vectorizer, and Classifier) pipeline, Paper IIII proposes a new alternative to it.
At first, the thesis introduces Paper I, an advanced log parser that fits into the parsing stage within the four-stage log anomaly detection system. The paper utilizes a rule-based tokenizer, interdependency token graph, and parallel processing to offer accurate and scalable log parsing. Moreover, the thesis dives deeper into log parsing with Paper II, introducing a new metric to measure character-level accuracy in log parsing alongside a new benchmark dataset tailored to that metric. This benchmark provides a more precise method of assessing and comparing log parsers' accuracy and performance.
For the last stage, the thesis introduces Paper III, which proposes a method built on the foundation of the Siamese networks. The paper differentiates itself from the competition by not only offering enhanced robustness against log evolution but also providing a solution to detect over-the-time data drifts in logs.
Additionally, Paper IIII replaces the parser, vectorizer, and classifier with a character-based Hierarchical Convolutional Neural Network (HCNN), transforming the multi-staged system into an end-to-end one. HCNN allows for analyzing the usually ignored digits, numbers, and punctuations alongside the natural language. Furthermore, Paper4 offers improved generalization across datasets, making it adaptable even in scenarios with limited training data.
Overall, the thesis contributes to software log anomaly detection in four papers. These contributions improve the system's accuracy, performance, scalability, and robustness across a diverse set of evaluations using public open datasets.Tiivistelmä
Ohjelmistolokit ovat puolistrukturoituja tekstitiedostoja, jotka sisältävät ohjelmiston suorituksen tapahtumat. Ne tarjoavat yksityiskohtaisia tietoja, jotka auttavat vikojen diagnosoinnissa, suorituskyvyn seurannassa ja tietoturvan varmistamisessa. Lokien monimutkaisuus ja määrä tekevät manuaalisesta lokien tarkastelusta epäkäytännöllistä, minkä vuoksi automaattinen lokianalyysi on välttämätöntä. Tämä väitöskirja hyödyntää useista eri lähteistä kerättyjä julkisia lokiaineistoja, jotka kattavat laajasti erilaisia ohjelmistojärjestelmiä ja supertietokoneita. Väitöskirja tarjoaa ratkaisuja ohjelmistolokien poikkeuksien tunnistukseen.
Opinnäyte esittelee Artikkelissa I kehittyneen lokijäsentimen, joka sijoittuu nelivaiheisen lokianomalian tunnistusjärjestelmän jäsentämisvaiheeseen. Artikkeli hyödyntää sääntöpohjaista tokenointia, tokenien välisiä riippuvuuksia mallintavaa graafia ja rinnakkaislaskentaa tarjotakseen tarkan ja skaalautuvan lokijäsentimen.
Seuraavaksi opinnäyte syventyy lokijäsentämiseen Artikkelissa II, joka esittelee uuden mittarin lokijäsentämisen tarkkuuden arviointiin merkkitasolla sekä uuden vertailuaineiston. Tämä vertailuaineisto tarjoaa tarkemman tavan arvioida ja vertailla lokijäsentimien tarkkuutta ja suorituskykyä.
Artikkeli III ehdottaa menetelmää poikkeusten luokitteluun, joka pohjautuu Siamese-neuroverkkoihin. Tämä lähestymistapa eroaa aiemmista tutkimuksista tarjoamalla paitsi parempaa sietokykyä lokien muutoksia kohtaan ja myös keinon havaita pitkän aikavälin datasiirtymiä lokeissa.
Lisäksi opinnäyte esittelee uudenlaisen lähestymistavan Artikkelissa IV. Tässä ratkaisussa jäsennin, vektorisointivaihe ja luokittelija korvataan merkkipohjaisella hierarkkisella konvoluutioneuroverkolla (HCNN), mikä muuntaa monivaiheisen järjestelmän päästä-päähän -ratkaisuksi. HCNN mahdollistaa usein huomiotta jäävien numeroiden ja välimerkkien analyysin yhdessä luonnollisen kielen kanssa. Artikkeli IV parantaa myös yleistettävyyttä eri aineistojen välillä, tehden siitä sovellettavan myös tilanteissa, joissa opetusaineistoa on niukasti. Kaiken kaikkiaan opinnäyte esittelee neljä menetelmää ohjelmistolokien analysointiin. Menetelmät parantavat poikkeuksien tunnistamisen tarkkuutta, suorituskykyä, skaalautuvuutta ja vikasietoisuutta laajassa joukossa julkisia lokiaineistoja
Effects of dimensional errors on compliant mechanisms performance by using dynamic splines
No full textThe paper deals with the description of a methodology for investigating the effects of the
dimensional errors on the performance of compliant assemblies in order to perform an optimal
allocation of tolerances. The proposed approach is based on the modeling of flexible elements
using the dynamic spline formulation which is able to take into account the nonlinear
compliance of the system using a small set of parameters. Starting from the description of the
elasto-kinematic behavior of the system written in terms of Lagrange equations, the deduction
of sensitivity coefficients is discussed and an algorithmic description of the tolerance allocation
scheme is presented. In order to give an example of application, the methodology has been
successfully applied to the optimization of a constant-force compliant mechanism
- …
