Abstract Recently, peptide-based drugs have gained unprecedented interest in discovering and developing antifungal due to their high efficacy, broad-spectrum activity, low toxicity few side effects. However, it is time-consuming expensive identify peptides (AFPs) experimentally. Therefore, computational methods for accurately predicting AFPs are highly required. In this work, we develop AFP-MFL, a novel deep learning model that predicts only relying on peptide sequences without using any...

Synthetic Aperture Radar (SAR) has been widely applied in geological exploration, military reconnaissance, and marine monitoring. However, due to hardware limitations, SAR images often suffer from low resolution high noise, making feature extraction challenging. To address this, we propose OS\_SR: Optical-SAR Feature Fusion based Image Super-resolution Reconstruction. This method enhances image reconstruction by integrating high-resolution high-frequency spatial features optical...

Alzheimer's disease (AD) endangers the physical and mental health of elderly, constituting one most crucial social challenges. Due to lack effective AD intervention drugs, it is very important diagnose in early stage, especially Mild Cognitive Impairment (MCI) phase.

Rumors spread rapidly through online social microblogs at a relatively low cost, causing substantial economic losses and negative consequences in our daily lives. Existing rumor detection models often neglect the underlying semantic coherence between text image components multimodal posts, as well challenges posed by incomplete modalities single modal such missing or images. This paper presents CLKD-IMRD, novel framework for Incomplete Modality Rumor Detection. CLKD-IMRD employs Contrastive...

Blood vessel segmentation is one of the important premises for analysis in Medical Imaging. In this paper, a new operation block called designed learning finer feature maps and keeping resolution invariant. Based on block, U-net-like model CDNet proposed blood task achieves precise results. The novel has been trained tested retinal dataset DRIVE achieved highest scores accuracy (0.9633), sensitivity (0.8158) AUC (0.9841) comparing with other methods. Our experimental results indicate that...

This article focuses on the task of Multi-Modal Summarization with Output for China JD.COM e-commerce product description containing both source text and images. In context learning multi-modal (text image) input, there exists a semantic gap between image, especially in cross-modal semantics image. As result, capturing shared earlier becomes crucial summarization. However, when generating summarization, based different contributions input images, relevance irrelevance contexts to target...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

The trenching device is an important part of the ecological grass truck. It can open trench, guide seeds and fertilizers into cover wet soil. Its structure design very important. An seeding cart combined plow type furrowing was designed, structural its furrow opener body, blade, bracket, connecting block other key components carried out. In order to further verify reasonable device, static analysis out by ANSYS Workbench software, stiffness strength met requirements use according...

Rod pumping systems are widely used in oil wells. Accurate fault prediction could reduce equipment rate and has practical significance improving oilfield production efficiency. This paper analyzed the journal of rod wells block X Xinjiang Oilfield. According to journal, well maintenance operations primarily caused by five types faults: scale, wax, corrosion, fatigue, wear. These faults make up approximately 90% all faults. 1354 that experienced workover as a result aforementioned factors...

Abstract With the application of personalized and precision medicine, more precise efficient antibody drug development technology is urgently needed. Identification antibody-antigen interactions crucial to engineering. The time-consuming expensive nature wet-lab experiments calls for computational methods. Taking into account non-overlapping advantage current structure-dependent sequence-only methods, we propose an interpretable interaction prediction method, S3AI. introduction structural...

One individual human’s immune repertoire consists of a huge set adaptive receptors at certain time point, representing the individual's state. Immune classification and associated receptor identification have potential to make transformative contribution development novel vaccines therapies. The vast number instances exceedingly low witness rate pose great challenge classification, which can be formulated as Massive Multiple Instance Learning (MMIL) problem. Traditional MIL methods, both...

With the rapid development and application of deep learning, its dataset size network model are becoming increasingly large, distributed training is popular. This article proposes a heterogeneous task scheduling resource allocation algorithm based on learning to address issues such as heterogeneity in usage, inability predict convergence time, communication time bottlenecks, waste caused by static during collaborative training. achieves dynamic tasks reduces completion clusters. The...

翻转课堂起源于美国，随着我国《大学英语教学指南》的出台，外语界正在积极探索实践翻转课堂模式，在提高学生自主学习能力的同时，也要充分满足学生个性化的学习期望。本文利用文献分析法，从翻转课堂教学指令前置、教学过程自主、教学主体转变这三个方面分析中国翻转课堂流变特征。同时，翻转课堂在中国嬗变中仍然存在隐形问题，即“翻而不转”“创而不新”。因此，本文尝试破解困境，根本目的是建构“先学后教、以学定教”的新型教学理念，教师需要高位指导、以学定教；学生需要保证权利、整合资源。

for improving the present situation in quality detection of plate making product line, a simplified homocentric square filter (SHSF) is proposed and implemented based on machine vision. Firstly, hardware platform including camera, lens, light source, encoder designed to acquire high images. In order automatically detect defects from acquired images, SHSF according characteristics their background. Furthermore, multiscale analysis applied different sizes. Combining with analysis, nine kinds...

The vision system of the mobile robot is a low-level function that provides required target information current environment for upper tasks. real-time performance and robustness object segmentation in cluttered environments still serious problem visions. In this paper, new indoor scene method based on RGB-D image, presented extracted primary regions are then used recognition. Firstly, paper accomplishes depth filtering by improved traditional method. Then using information, algorithm...

In this paper, a new convolutional neural network called multi-U fusion networks (MUFNet) is proposed for accurate semantic segmentation of multi-spectral remote sensing. Essentially, MUFNet inspired by UNet, MFNet and CAM fully combines their advantages. First, introduces the skip connections into multi-encoder-to-mono-decoder architecture, thereby facilitating multi-scale multi-channel spectral information. Second, shortcut module in decoder revised concatenating multiple features from...

Synthetic Aperture Radar (SAR) image super-resolution (SR) has broad application prospects in military reconnaissance, crop monitoring and other fields. However, current SAR SR limitations like insufficient feature extraction, difficulty of representation, modeling reconstruction. Therefore, this paper adopts high-resolution optical images as auxiliary to improve the performance SR. We propose an Optical Guidance Residual Networks (OGRN) based on residual learning for Firstly, we extract...

Due to the rapid development of mobile robots technology, object recognition is great practical significance. The real-time performance and robustness segmentation in cluttered environments a considerable problem robot vision. In this paper, new method using depth information presented. Firstly, approach obtains candidate region clue, then accomplished filtering region. Next, extended get better edge information. Finally, foreground extracted results realized on color image. This was tested...