Acquiring labels in anomaly detection tasks is expensive and challenging. Therefore, as an effective way to improve efficiency, pretraining is widely used in anomaly detection models, which enriches the model's representation capabilities, thereby enhancing both performance and efficiency in anomaly detection. In most pretraining methods, the decoder is typically randomly initialized. Drawing inspiration from the diffusion model, this paper proposed to use denoising as a task to pretrain the decoder in anomaly detection, which is trained to reconstruct the original noise-free input. Denoising requires the model to learn the structure, patterns, and related features of the data, particularly when training samples are limited. This paper explored two approaches on anomaly detection: simultaneous denoising pretraining for encoder and decoder, denoising pretraining for only decoder. Experimental results demonstrate the effectiveness of this method on improving model’s performance. Particularly, when the number of samples is limited, the improvement is more pronounced.
2
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Marine aerosol, containing an enormous source of chloride, coupled with severe environmental conditions (e.g., high temperature, high relative humidity), poses a threat to the durability of concrete exposed to the marine atmospheric zone. The distribution of marine aerosol is spatial and temporal dependent, and thus, the deposition rate of airborne chlorides Ddep can vary a lot with geological and environmental factors. Chloride profile in concrete exposed to marine aerosol is a two-zone profile due to the wetting/drying action. The peak chloride concentration Cmax and depth of the convection zone Δx are largely affected by time, materials, environmental conditions which usually is less than 10 mm. Many models based on Fick’s law are developed to predict chloride transport in unsaturated concrete under wetting-drying cycles. However, the prediction of marine aerosol penetration into concrete is far from satisfactory, due to lack of enough experimental and theoretical researches.
3
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Sodium and cerium co-doped Bi4Ti2.86W0.14O12 ceramics abided by the formula (Na0.5Ce0.5)xBi4-xTi2.86W0.14O12 (abbreviated as NC100x-BITW, x = 0-0.12) were prepared by solid state reaction method. The effect of (Na0.5Ce0.5) addition on the structure and electrical properties of the ceramics was investigated. XRD results indicated that the NC100x-BITW ceramics possess a pure three-layer Aurivillius-type structure. (Na0.5Ce0.5) addition first increases and then decreases the grain size which can be observed by scanning electron microscopy. With the increase of (Na0.5Ce0.5) addition, the Curie temperature (Tc) was gradually decreased from 632 C to 595 C. The piezoelectric properties can be enhanced while the dielectric loss decreased after (Na0.5Ce0.5) addition, and the optimal properties were obtained as follows when x = 0.06: d33 = 21.1 pC/N, kp = 7.2%, Qm = 4978, εr =147 (@100 kHz), tanδ = 0.27% (@100 kHz), Tc =614 C and resistivity of 4.3•108 Ω•cm at 500 C, suitable for high temperature sensing applications.
PL
Ceramikę Bi4Ti2.86W0.14O12 domieszkowaną sodem i cerem wytworzono zgodnie ze wzorem (Na0.5Ce0.5)xBi4-xTi2.86W0.14O12 (w skrócie NC100x-BITW, x = 0-0,12) metodą reakcji w fazie stałej. Zbadano wpływ dodatku (Na0,5Ce0,5) na budowę i właściwości elektryczne ceramiki. Wyniki XRD wykazały, że ceramika NC100x-BITW ma czystą, trójwarstwową strukturę typu Aurivilliusa. Dodanie (Na0,5Ce0,5) najpierw zwiększa, a następnie zmniejsza rozmiar ziarna, co można zaobserwować za pomocą skaningowej mikroskopii elektronowej. Wraz ze wzrostem dodatku (Na0,5Ce0,5) temperatura Curie (Tc) stopniowo obniżała się z 632 C do 595 C. Właściwości piezoelektryczne można poprawić, podczas gdy stratę dielektryczną zmniejszyć po dodaniu (Na0,5Ce0,5) i następujące optymalne właściwości uzyskano, gdy x = 0,06: d33 = 21,1 pC/N, kp = 7,2%, Qm = 4978, εr = 147 (@ 100 kHz), tgδ = 0,27% (@ 100 kHz), Tc = 614 C i rezystywność 4,3•108 Ω•cm przy 500 C, odpowiednie do zastosowań z czujnikami wysokotemperaturowymi.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.