1961. |
1962. |
1963. |
1964. |
1965. |
1966. |
1967. A cross-benchmark examination of feature-based algorithm selector generalization in single-objective numerical optimizationGjorgjina Cenikj, Gašper Petelin, Tome Eftimov, 2024, original scientific article Abstract: The task of selecting the best optimization algorithm for a particular problem is known as algorithm selection (AS). This involves training a model using landscape characteristics to predict algorithm performance, but a key challenge remains: making AS models generalize effectively to new, untrained benchmark suites. This study assesses AS models’ generalizability in single-objective numerical optimization across diverse benchmark suites. Using Exploratory Landscape Analysis (ELA) and transformer-based (TransOpt) features, the research investigates their individual and combined effectiveness in AS across four benchmarks: BBOB, AFFINE, RANDOM, and ZIGZAG. AS models perform differently based on benchmark suite similarities in algorithm performance distributions and single-best solvers. When suites align, these models underperform against a baseline predicting mean algorithm performance; yet, they outperform the baseline when suites differ in performance distributions and solvers. The AS models trained using the ELA landscape features are better than the models trained using the TransOpt features on the BBOB and AFFINE benchmark suites, while the opposite is true for the RANDOM benchmark suite. Ultimately, the study reveals challenges in accurately capturing algorithm performance through problem landscape features (ELA or TransOpt), impacting AS model applicability.
Keywords: algorithm selection, multi-target regression, generalization, benchmarking
Published in DiRROS: 21.05.2024; Views: 528; Downloads: 532
 Full text (2,49 MB)
This document has many files! More... |
1968. |
1969. |
1970. Izdelava sestojne karta na podlagi lidarskih podatkovAndrej Bončina, Vasilije Trifković, Christian Rosset, 2024, original scientific article Abstract: Sestojna karta je pomemben vir podatkov o gozdnih sestojih. Postopki izdelave sestojne karte so različni. V prispevku prikazujemo i) postopek izdelave avtomatizirane sestojne karte (TBk), ki so ga razvili na Bern University of Applied Sciences (BFH), ii) izdelani sestojni karti TBk za gozdni območji Draga in Pokljuka, iii) klasifikacijo sestojev v sestojne tipe na teh območjih ter iv) rezultate primerjave avtomatske sestojne karte s sestojno karto, ki jo izdeluje Zavod za gozdove Slovenije. Pri izdelavi TBk je dominanta višina drevja temeljni kriteriji razmejevanja sestojev, dominantna višina in stopnje zastiranje po sestojnih plasteh pa so temeljni kriteriji za klasifikacijo sestojev v sestojne tipe. Na TBk karti v Dragi in na Pokljuki je povprečna velikost sestoja 1,76 ha in 1,64 ha , kar je precej manj od povprečne velikosti sestoja na karti ZGS (5,90 ha in 2,85 ha). V velikostni strukturi sestojev na TBk v skupnem številu sestojev prevladujejo sestoji s površino, manjšo od 0,5 ha. Deleži sestojnih tipov na TBk in karti ZGS se ujemajo, razmejitev sestojev pa se razlikuje zaradi različnih kriterijev in podrobnejše obravnave sestojev pri TBk. Aktualnost karte TBk je pogojena s starostjo lidarskih posnetkov. Prednosti TBk so hitra izdelava, objektivnost razmejevanja, možnost spremljanja sprememb gozdnih sestojev v času, večplastne informacije o sestojih, poglavitna omejitev pa je omejen nabor sestojnih znakov, kot so poškodovanost, negovanost, kakovost, ki jih sicer pridobimo pri opisovanju gozdov, TBk tudi ne obsega podrobnega načrta.
Keywords: sestojna karta, lidarski podatki, klasifikacija sestojev, TBk
Published in DiRROS: 21.05.2024; Views: 550; Downloads: 215
Full text (614,02 KB) |
