128,900 research outputs found
A Subband-Selective Broadband GSC with Cosine-Modulated Blocking Matrix
No full textIn this paper, a novel subband-selective generalized sidelobe canceller (GSC) for partially adaptive broadband beamforming is proposed. The columns of the blocking matrix are derived from a prototype vector by cosine-modulation, and the broadside constraint is incorporated by imposing zeros on the prototype vector appropriately. These columns constitute a series of bandpass filters, which select signals with specific angles of arrival and frequencies. This results in highpass-type bandlimited spectra of the blocking matrix outputs, which is further exploited by subbands decomposition and suitably discarding the low-pass subbands prior to running independent unconstrained adaptive filters in each non-redundant subband. By these steps, the computational complexity of a GSC implementation is greatly reduced compared to fully adaptive GSC schemes, while performance is comparable or even enhanced due to subband decorrelation in both spatial and temporal domains
Difference based Ridge and Liu type Estimators in Semiparametric Regression Models
No full textWe consider a difference based ridge regression estimator and a Liu type estimator of the regression parameters in the partial linear semiparametric regression model, y = Xβ + f + ε. Both estimators are analysed and compared in the sense of mean-squared error. We consider the case of independent errors with equal variance and give conditions under which the proposed estimators are superior to the unbiased difference based estimation technique. We extend the results to account for heteroscedasticity and autocovariance in the error terms. Finally, we illustrate the performance of these estimators with an application to the determinants of electricity consumption in Germany.Difference based estimator; Differencing estimator, Differencing matrix, Liu estimator, Liu type estimator, Multicollinearity, Ridge regression estimator, Semiparametric model
The extraction and anti-inflammatory screening of <i>Onosma glomeratum</i> Y. L. Liu
No full text“Zicao” has a long medicinal history and has a variety of pharmacological activities. As the main resource of “zicao” in Tibet, Onosma glomeratum Y. L. Liu (tuan hua dian zi cao), usually used for treating pneumonia in Tibet, has not been reported deeply. In order to determine the main anti-inflammatory active ingredients of Onosma glomeratum Y. L. Liu, in this study, the extracts enriched in naphthoquinones and polysaccharides were optimized prepared form Onosma glomeratum Y. L. Liu by ultrasonic extraction, and reflux extraction, respectively, with Box-Behnken design effect surface method. And their anti-inflammatory abilities were screened on LPS induced A549 cells model, for figuring out the anti-inflammatory active ingredients from Onosma glomeratum Y. L. Liu. The extract enriched naphthoquinone was obtained under following condition: extract with 85% ethanol in a liquid to material ratio of 1:40 g/mL at 30 °C for 30 minutes using ultrasound, leading to the extraction rate of total naphthoquinone as 0.98 ± 0.017%; the extract enriched polysaccharides was prepared as follows: extract 82 minutes at 100 °C with distilled water in a liquid to material ratio of 1:50 g/mL, with extraction rate of polysaccharide as 7.07 ± 0.02%. On the LPS-induced A549 cell model, the polysaccharide extract from Onosma glomeratum Y. L. Liu showed better anti-inflammatory effects than the naphthoquinone extract, indicating the extract enriched in polysaccharides is the anti-inflammatory extract of Onosma glomeratum Y. L. Liu, which could serve as a potential anti-inflammatory extract in medical and food industries in the future.</p
Toward scalable and unbiased scene graph generation : active learning and causal inference perspectives
No full textAbstract
Scene Graph Generation (SGG) aims to construct structured representations of visual scenes by identifying objects and their pairwise relationships. Despite significant progress, SGG remains challenged by two core issues: the high cost of triplet-level annotations and the persistent biases in relationship prediction. This thesis addresses both challenges through two complementary research threads—label-efficient SGG and bias-mitigated SGG.
In the first thread, we introduce EDAL, a novel active learning framework that integrates evidential uncertainty estimation with diversity-aware sample selection. EDAL enables SGG models to achieve competitive performance using only 10\% of labeled data, dramatically reducing annotation costs while preserving generalization. In the second thread, we investigate the origin of biased predictions in SGG, which are often caused by long-tailed distributions, semantic confusion, and spurious correlations. To this end, we present a series of causal debiasing methods—TsCM, CAModule, and RcSGG—that leverage structural causal modeling, triplet-level logit adjustment, and reverse causal reasoning to systematically mitigate these biases.
Our methods are validated on multiple SGG benchmarks and backbones, achieving state-of-the-art performance on debiasing metrics while preserving overall accuracy. In addition, we highlight the potential of large-scale foundation models in future SGG research, given their success in improving generalization and alleviating annotation burdens in other vision tasks.
By unifying active learning and causal debiasing, this thesis offers a comprehensive framework for building scalable and fair SGG systems, opening new directions for structured visual understanding. Original papers Sun, S., Zhi, S., Heikkilä, J., & Liu, L. (2023). Evidential uncertainty and diversity guided active learning for scene graph generation. In ICLR 2023 : The Eleventh International Conference on Learning Representations. OpenReview.net. https://openreview.net/forum?id=xI1ZTtVOtlz Self-archived version Sun, S., Zhi, S., Liao, Q., Heikkilä, J., & Liu, L. (2023). Unbiased scene graph generation via two-stage causal modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(10), 12562–12580. https://doi.org/10.1109/TPAMI.2023.3285009 https://doi.org/10.1109/TPAMI.2023.3285009 Self-archived version Liu, L., Sun, S., Zhi, S., Shi, F., Liu, Z., Heikkilä, J., & Liu, Y. (2025). A causal adjustment module for debiasing scene graph generation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(5), 4024–4043. https://doi.org/10.1109/TPAMI.2025.3537283 https://doi.org/10.1109/TPAMI.2025.3537283 Self-archived version Sun, S., Liu, L., Liu, T., Zhi, S., Cheng, M.-M., Heikkilä, J., & Liu, Y. (2025). A reverse causal framework to mitigate spurious correlations for debiasing scene graph generation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(9), 7470–7489. https://doi.org/10.1109/TPAMI.2025.3568644 https://doi.org/10.1109/TPAMI.2025.3568644 Self-archived version Tiivistelmä
Scene Graph Generation (SGG) pyrkii rakentamaan visuaalisista kohtauksista jäsenneltyjä esityksiä tunnistamalla objektit ja niiden väliset parisuhteet. Huolimatta merkittävästä edistyksestä alalla, SGG:tä vaivaavat edelleen kaksi keskeistä haastetta: kolmikkoanotointien korkeat kustannukset sekä pysyvät vinoumat suhdepäätelmissä. Tämä väitöskirja käsittelee molempia ongelmia kahden toisiaan täydentävän tutkimuslinjan kautta: tehokas annotointi (label-efficient SGG) ja vinoumien poistaminen (bias-mitigated SGG).
Ensimmäisessä tutkimuslinjassa esittelemme EDAL:n, uuden aktiiviseen oppimiseen perustuvan kehyksen, joka yhdistää todisteperusteisen epävarmuuden arvioinnin sekä monimuotoisuustietoisen näytevalinnan. EDAL mahdollistaa SGG-mallien kilpailukykyisen suorituskyvyn käyttämällä vain 10\% anotetuista tiedoista, mikä vähentää merkittävästi anotointikustannuksia ilman, että yleistettävyys kärsii. Toisessa tutkimuslinjassa tarkastelemme SGG:n vinoutuneiden ennusteiden alkuperää, jotka johtuvat usein pitkähäntäisistä jakaumista, semanttisesta sekaannuksesta ja näennäiskorrelaatioista. Esittelemme joukon kausaaliperustaisia vinoumanpoistomenetelmiä—TsCM, CAModule ja RcSGG—jotka hyödyntävät rakenteellista kausaalimallinnusta, triplatasoista logit-säätöä ja käänteistä kausaalipäättelyä näiden vinoumien systemaattiseen lievittämiseen.
Menetelmämme on validoitu useilla SGG-vertailuaineistoilla ja selkärankamalleilla, ja ne saavuttavat huipputason tuloksia vinoumametriikoissa samalla säilyttäen kokonaistarkkuuden. Lisäksi tuomme esiin suurten perusmallien (foundation models) potentiaalin tulevassa SGG-tutkimuksessa, erityisesti niiden yleistämiskyvyn ja anotointitarpeen keventämisen ansiosta muissa visuaalisissa tehtävissä.
Yhdistämällä aktiivisen oppimisen ja kausaalisen vinoumanpoiston tämä väitöskirja esittää kokonaisvaltaisen kehyksen skaalautuvien ja reilujen SGG-järjestelmien rakentamiseen, avaten uusia tutkimussuuntia jäsenneltyyn visuaaliseen ymmärrykseen. Osajulkaisut Sun, S., Zhi, S., Heikkilä, J., & Liu, L. (2023). Evidential uncertainty and diversity guided active learning for scene graph generation. In ICLR 2023 : The Eleventh International Conference on Learning Representations. OpenReview.net. https://openreview.net/forum?id=xI1ZTtVOtlz Rinnakkaistallennettu versio Sun, S., Zhi, S., Liao, Q., Heikkilä, J., & Liu, L. (2023). Unbiased scene graph generation via two-stage causal modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(10), 12562–12580. https://doi.org/10.1109/TPAMI.2023.3285009 https://doi.org/10.1109/TPAMI.2023.3285009 Rinnakkaistallennettu versio Liu, L., Sun, S., Zhi, S., Shi, F., Liu, Z., Heikkilä, J., & Liu, Y. (2025). A causal adjustment module for debiasing scene graph generation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(5), 4024–4043. https://doi.org/10.1109/TPAMI.2025.3537283 https://doi.org/10.1109/TPAMI.2025.3537283 Rinnakkaistallennettu versio Sun, S., Liu, L., Liu, T., Zhi, S., Cheng, M.-M., Heikkilä, J., & Liu, Y. (2025). A reverse causal framework to mitigate spurious correlations for debiasing scene graph generation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(9), 7470–7489. https://doi.org/10.1109/TPAMI.2025.3568644 https://doi.org/10.1109/TPAMI.2025.3568644 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 via remote access, on 27 November 2025, at 9 a.m.Abstract
Scene Graph Generation (SGG) aims to construct structured representations of visual scenes by identifying objects and their pairwise relationships. Despite significant progress, SGG remains challenged by two core issues: the high cost of triplet-level annotations and the persistent biases in relationship prediction. This thesis addresses both challenges through two complementary research threads—label-efficient SGG and bias-mitigated SGG.
In the first thread, we introduce EDAL, a novel active learning framework that integrates evidential uncertainty estimation with diversity-aware sample selection. EDAL enables SGG models to achieve competitive performance using only 10\% of labeled data, dramatically reducing annotation costs while preserving generalization. In the second thread, we investigate the origin of biased predictions in SGG, which are often caused by long-tailed distributions, semantic confusion, and spurious correlations. To this end, we present a series of causal debiasing methods—TsCM, CAModule, and RcSGG—that leverage structural causal modeling, triplet-level logit adjustment, and reverse causal reasoning to systematically mitigate these biases.
Our methods are validated on multiple SGG benchmarks and backbones, achieving state-of-the-art performance on debiasing metrics while preserving overall accuracy. In addition, we highlight the potential of large-scale foundation models in future SGG research, given their success in improving generalization and alleviating annotation burdens in other vision tasks.
By unifying active learning and causal debiasing, this thesis offers a comprehensive framework for building scalable and fair SGG systems, opening new directions for structured visual understanding.Tiivistelmä
Scene Graph Generation (SGG) pyrkii rakentamaan visuaalisista kohtauksista jäsenneltyjä esityksiä tunnistamalla objektit ja niiden väliset parisuhteet. Huolimatta merkittävästä edistyksestä alalla, SGG:tä vaivaavat edelleen kaksi keskeistä haastetta: kolmikkoanotointien korkeat kustannukset sekä pysyvät vinoumat suhdepäätelmissä. Tämä väitöskirja käsittelee molempia ongelmia kahden toisiaan täydentävän tutkimuslinjan kautta: tehokas annotointi (label-efficient SGG) ja vinoumien poistaminen (bias-mitigated SGG).
Ensimmäisessä tutkimuslinjassa esittelemme EDAL:n, uuden aktiiviseen oppimiseen perustuvan kehyksen, joka yhdistää todisteperusteisen epävarmuuden arvioinnin sekä monimuotoisuustietoisen näytevalinnan. EDAL mahdollistaa SGG-mallien kilpailukykyisen suorituskyvyn käyttämällä vain 10\% anotetuista tiedoista, mikä vähentää merkittävästi anotointikustannuksia ilman, että yleistettävyys kärsii. Toisessa tutkimuslinjassa tarkastelemme SGG:n vinoutuneiden ennusteiden alkuperää, jotka johtuvat usein pitkähäntäisistä jakaumista, semanttisesta sekaannuksesta ja näennäiskorrelaatioista. Esittelemme joukon kausaaliperustaisia vinoumanpoistomenetelmiä—TsCM, CAModule ja RcSGG—jotka hyödyntävät rakenteellista kausaalimallinnusta, triplatasoista logit-säätöä ja käänteistä kausaalipäättelyä näiden vinoumien systemaattiseen lievittämiseen.
Menetelmämme on validoitu useilla SGG-vertailuaineistoilla ja selkärankamalleilla, ja ne saavuttavat huipputason tuloksia vinoumametriikoissa samalla säilyttäen kokonaistarkkuuden. Lisäksi tuomme esiin suurten perusmallien (foundation models) potentiaalin tulevassa SGG-tutkimuksessa, erityisesti niiden yleistämiskyvyn ja anotointitarpeen keventämisen ansiosta muissa visuaalisissa tehtävissä.
Yhdistämällä aktiivisen oppimisen ja kausaalisen vinoumanpoiston tämä väitöskirja esittää kokonaisvaltaisen kehyksen skaalautuvien ja reilujen SGG-järjestelmien rakentamiseen, avaten uusia tutkimussuuntia jäsenneltyyn visuaaliseen ymmärrykseen
Amplitude-Phase Detection for Power Converters Tied to Unbalanced Grids with Large X/R Ratios
No full textAmplitude-phase information of the voltage/current phasor, which is the prerequisite for grid-tied converter control and protection, can change almost instantaneously in low-voltage distribution systems with small X/R ratios. However, the equivalent impedances of high-voltage power systems are usually inductive (with large X/R ratios), and the power flow of pertinent grids generates decaying dc (DDC) components with large amplitude and long duration during large disturbances. Therefore, the X/R ratios and DDC components must be fully considered in amplitude-phase detection to achieve effective converter control and protection. To this end, the main components present in the transient voltage/current of a high-voltage power system after unbalanced disturbance are first analyzed, and the general transient model is obtained by synthesizing multimode DDC components. Then, by resorting to waveform characteristics of different components, an amplitude-phase detection algorithm is proposed based on the multicomponent parallel-detection structure. Compared to existing techniques, the detection time is reduced to half grid cycle. Finally, an iterative variable-interval integral algorithm is developed, improving the antinoise ability of detection algorithm and overcoming the computational burden issue. This enables the direct integration of algorithm into the converter's embedded processor. Experiment results verified the effectiveness and superiority of the proposed technique
Frequency Trajectory Planning Based Strategy for Improving Frequency Stability of Droop-Controlled Inverter Based Standalone Power Systems
No full textInverter-based standalone power systems (ISPSs), with limited inertia/damping, are prone to significant changes in frequency indicators encompassing frequency deviation and rate of change of frequency (RoCoF), which can easily exceed the pertinent relay thresholds and cause power outages. Limited by estimating the disturbance and system inertia/damping deficiency, existing controls cannot ensure system frequency stability. To overcome this issue, a frequency trajectory planning (FTP) based strategy is developed in this article to improve frequency stability of droop-controlled ISPS. This frequency-indicator-oriented control relates system frequency with several pre-defined planning parameters, decoupling from system's disturbances and inertia/damping deficiency that are difficult to evaluate. During normal operation or when the disturbance level is insignificant, the inverter operates in standard droop mode since the system intrinsic inertia/damping features suffice to regulate the frequency. In the event of a large disturbance, an FTP block is triggered by detected frequency indicators, and a marginally safe frequency trajectory is planned according to the requirement of the grid code. In this case, by tracking the planned frequency trajectory, the inverter provides suitable inertia/damping support needed by the system, thereby guaranteeing frequency stability of the ISPS. Finally, simulation and experimental results proved the effectiveness and advancement of the FTP based strategy
Frequency Trajectory Planning Based VSC Control Strategy for Power System Frequency Regulation
No full textTo ensure frequency stability of low inertia power systems, VSC needs to provide frequency response (FR) service. Limited by predicting grid inertia/damping and disturbances that are difficult to measure, the FR performance is greatly restricted. Besides, excessive response under certain conditions results in several consequences, including 1) energy waste, and 2) the VSC power frequently and excessively deviating from steady-state points, thereby reducing the component's life, lowering system safety/reliability, and increasing the risk of VSC overcurrent faults. To solve these issues, this paper proposes the frequency trajectory planning (FTP) based FR strategy, with the fundamental idea to actively plan the frequency trajectory with necessary safety margin in accordance with the grid code. Specifically, if the system frequency locates within the range of planned one (suggesting frequency indexes satisfying the requirement), VSC remains at the steady-state point to avoid excessive FR service; otherwise, the FR service is required, and the FTP control is enabled to closely track the planned trajectory. By fulfilling the tracking purpose, the requirement on system frequency indexes is met, frequency stability is guaranteed, and the over-response is avoided. Namely, the planned trajectory can characterize whether the system requires FR service and whether the provided FR service is sufficient or excessive. Additionally, the method needs no measurement of inertia/damping and disturbances of the system, and can be directly integrated into the commonly used VSC control systems. Finally, experiments prove the effectiveness and advancement of the FTP strategy
Insights into the antibacterial mechanism of PEGylated nano-bacitracin A against Streptococcus pneumonia: both penicillin-sensitive and penicillin-resistant strains [Corrigendum]
No full textInsights into the antibacterial mechanism of PEG ylated nano-bacitracin A against Streptococcus pneumonia: both penicillin-sensitive and penicillinresistant strains [Corrigendum]Hong W, Liu L, Zhang Z, Zhao Y, Zhang D, Liu M. Int J Nanomedicine. 2018;13:6297–6309.The author wishes to advise that on page 6306, Figure 7 parts A and B are incorrect. The correct figure parts are included below:Read the original articl
- …
