Model-based methods for recommender systems have been studied extensively for years. Modern recommender systems usually resort to 1) representation learning models which define user-item preference as the distance between their embedding representations, and 2) embedding-based Approximate Nearest Neighbor (ANN) search to tackle the efficiency problem introduced by large-scale corpus. While providing efficient retrieval, the embedding-based retrieval pattern also limits the model capacity since the form of user-item preference measure is restricted to the distance between their embedding representations. However, for other more precise user-item preference measures, e.g., preference scores directly derived from a deep neural network, they are computationally intractable because of the lack of an efficient retrieval method, and an exhaustive search for all user-item pairs is impractical. In this paper, we propose a novel method to extend ANN search to arbitrary matching functions, e.g., a deep neural network. Our main idea is to perform a greedy walk with a matching function in a similarity graph constructed from all items. To solve the problem that the similarity measures of graph construction and user-item matching function are heterogeneous, we propose a pluggable adversarial training task to ensure the graph search with arbitrary matching function can achieve fairly high precision. Experimental results in both open source and industry datasets demonstrate the effectiveness of our method. The proposed method has been fully deployed in the Taobao display advertising platform and brings a considerable advertising revenue increase. We also summarize our detailed experiences in deployment in this paper.
Piezoelectric catalysis could convert mechanical energy into chemical energy, which can combine with solar energy for a high-efficiency piezo-photocatalysis reaction. In this work, NiTiO
Infants exhibit remarkable language acquisition abilities, supported by highly plastic neural substrates that dynamically interact with early speech experiences. However, the developmental mechanisms of these neural substrates and their specific role in speech acquisition remain incompletely understood. Here, we present NeoAudi Tract (NAT), a robust automated toolbox for extracting the full set of auditory tracts in infants from birth to 24 months using $3$T diffusion MRI data. By characterizing the microstructural changes in these tracts, we demonstrate a gradual and continuous maturation process of the auditory system. Additionally, we identify significant correlations between auditory tract maturation and both \textit{fine-motor skills} and \textit{expressive language} $t$-scores from the Mullen Scales of Early Learning tests. Our findings highlight the role of the auditory system in speech production and indicate the intertwined development of auditory and motor systems that underlies speech acquisition, particularly during perceptual reorganization.
Objective
To study the pathogenesis, diagnosis and treatment of pancreatic portal hypertension (PPH).
Methods
The clinical data of 37 patients with PPH treated in Henan Province People's Hospital from January 2008 to January 2016 were retrospectively analyzed.
Result
Nine patients underwent conservative treatment and 28 patients underwent surgical treatment. No deaths were observed in the perio-perative and follow-up periods. One patient underwent a second operation because of gastrointestinal bleeding. The clinical symptoms of the remaining patients were significantly relieved after surgery.
Conclusions
Treatment should be individualized and directed at the underlying cause. The anatomy of the coronary vein and the location of obstruction of the splenic vein determined the degree of the variceal veins and the surgical methods. Splenectomy was the basic treatment for PPH. Subcapsular splenectomy was effective in some challenging cases.
Key words:
Pancreatic portal hypertension; Splenectomy
Model-based methods for recommender systems have been studied extensively for years. Modern recommender systems usually resort to 1) representation learning models which define user-item preference as the distance between their embedding representations, and 2) embedding-based Approximate Nearest Neighbor (ANN) search to tackle the efficiency problem introduced by large-scale corpus. While providing efficient retrieval, the embedding-based retrieval pattern also limits the model capacity since the form of user-item preference measure is restricted to the distance between their embedding representations. However, for other more precise user-item preference measures, e.g., preference scores directly derived from a deep neural network, they are computationally intractable because of the lack of an efficient retrieval method, and an exhaustive search for all user-item pairs is impractical.
Considering the issues related to significant fluctuations and errors in the measured data of the dual-energy gamma-ray online ash analyzer caused by various external factors, this study investigates the impact of factors such as the distance between the detector and the radioactive source, coal seam thickness, moisture content, and particle size range on ash measurements. Additionally, a novel low-energy gamma-ray online ash measurement device is proposed, which ensures uniform coal seam thickness. By utilizing this device, the coal samples subjected to measurement exhibit characteristics such as small particle size, a narrow range of particle sizes, a smooth coal flow surface, consistent coal flow density, a fixed distance between the coal seam and the detector surface, and constant coal seam thickness. These conditions collectively provide optimal circumstances for accurate low-energy gamma-ray ash content measurement. Experimental results indicate that during the separation of Suntuan: Tongting raw coal, the average discrepancy between the measured ash content and the assay ash content was only 0.33%, and the standard deviation of the measured ash content was significantly lower than that of the assay ash content. Specifically, when separating Suntuan: Tongting raw coal, the standard deviation reached its minimum at 0.16%. Comparatively, the low-energy gamma-ray online ash measurement device outperforms the dual-energy gamma-ray online ash analyzer, exhibiting enhanced accuracy and stability in ash measurements, thereby better fulfilling the requirements of daily production.
Profound congenital sensorineural hearing loss (SNHL) prevents children from developing spoken language. Cochlear implantation and auditory brainstem implantation can provide partial hearing sensation, but language development outcomes can vary, particularly for patients with inner ear malformations and/or cochlear nerve deficiency (IEM&CND). Currently, the peripheral auditory structure is evaluated through visual inspection of clinical imaging, but this method is insufficient for surgical planning and prognosis. The central auditory pathway is also challenging to examine in vivo due to its delicate subcortical structures. Previous attempts to locate subcortical auditory nuclei using fMRI responses to sounds are not applicable to patients with profound hearing loss as no auditory brainstem responses can be detected in these individuals, making it impossible to capture corresponding blood oxygen signals in fMRI. In this study, we developed a new pipeline for mapping the auditory pathway using structural and diffusional MRI. We used a fixel-based approach to investigate the structural development of the auditory-language network for profound SNHL children with normal peripheral structure and those with IEM&CND under 6 years old. Our findings indicate that the language pathway is more sensitive to peripheral auditory condition than the central auditory pathway, highlighting the importance of early intervention for profound SNHL children to provide timely speech inputs. We also propose a comprehensive pre-surgical evaluation extending from the cochlea to the auditory-language network, showing significant correlations between age, gender, Cn.VIII median contrast value, and the language network with post-implant qualitative outcomes.
In this paper, an automatic colony counting system based on an improved image preprocessing algorithm and convolutional neural network (CNN)-assisted automatic counting method was developed. Firstly, we assembled an LED backlighting illumination platform as an image capturing system to obtain photographs of laboratory cultures. Consequently, a dataset was introduced consisting of 390 photos of agar plate cultures, which included 8 microorganisms. Secondly, we implemented a new algorithm for image preprocessing based on light intensity correction, which facilitated clearer differentiation between colony and media areas. Thirdly, a U2-Net was used to predict the probability distribution of the edge of the Petri dish in images to locate region of interest (ROI), and then threshold segmentation was applied to separate it. This U2-Net achieved an F1 score of 99.5% and a mean absolute error (MAE) of 0.0033 on the validation set. Then, another U2-Net was used to separate the colony region within the ROI. This U2-Net achieved an F1 score of 96.5% and an MAE of 0.005 on the validation set. After that, the colony area was segmented into multiple components containing single or adhesive colonies. Finally, the colony components (CC) were innovatively rotated and the image crops were resized as the input (with 14,921 image crops in the training set and 4281 image crops in the validation set) for the ResNet50 network to automatically count the number of colonies. Our method achieved an overall recovery of 97.82% for colony counting and exhibited excellent performance in adhesion classification. To the best of our knowledge, the proposed “light intensity correction-based image preprocessing→U2-Net segmentation for Petri dish edge→U2-Net segmentation for colony region→ResNet50-based counting” scheme represents a new attempt and demonstrates a high degree of automation and accuracy in recognizing and counting single-colony and multi-colony targets.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
