MixIR: Mixing Input and Representations for Contrastive Learning.

Recently, contrastive learning has shown significant progress in learning visual representations from unlabeled data. The core idea is training the backbone to be invariant to different augmentations of an instance. While most methods only maximize the feature similarity between two augmented data, we further generate more challenging training samples and force the model to keep predicting aggregated representation on these hard samples.

Propagation properties of elegant modified Bessel Gaussian beams.

A kind of optical beam with a radially parabolic propagating manner and intensity decay inversely proportional to propagating distance in the far field is investigated. The initial complex amplitudes of this kind of beam have the form of a Gaussian function multiplied by a /2-order modified Bessel function and a helical phase factor with topological charge . The arguments for Bessel and Gauss parts in the propagating solutions of these beams are complex and symmetric as elegant Laguerre and Hermite Gaussian beams.

Superhydrophilic Modification of Polycarbonate Substrate Surface by Organic Plasma Polymerization Film.

Superhydrophilicity performs well in anti-fog and self-cleaning applications. In this study, polycarbonate substrate was used as the modification object because of the low surface energy characteristics of plastics. Procedures that employ plasma bombardment, such as etching and high surface free energy coating, are applied to improve the hydrophilicity.

Unsupervised Person Re-Identification With Stochastic Training Strategy.

Unsupervised person re-identification (re-ID) has attracted increasing research interests because of its scalability and possibility for real-world applications. State-of-the-art unsupervised re-ID methods usually follow a clustering-based strategy, which generates pseudo labels by clustering and maintains a memory to store instance features and represent the centroid of the clusters for contrastive learning. This approach suffers two problems.

Predicting the Risk of Postoperative Complications of Schwannoma Surgery: Development and Assessment of a New Predictive Nomogram.

Background:  The aim of this study was to develop and internally validate a risk nomogram for postoperative complications of schwannoma surgery.

Methods:  From 2016 to 2020, we reviewed 83 patients who underwent schwannoma resection with a total number of 85 schwannomas. A predictive model was developed based on the dataset of this group.

Delivery of pOXR1 through an injectable liposomal nanoparticle enhances spinal cord injury regeneration by alleviating oxidative stress.

Oxidation resistance 1 (OXR1) is regarded as a critical regulator of cellular homeostasis in response to oxidative stress. However, the role of OXR1 in the neuronal response to spinal cord injury (SCI) remains undefined. On the other hand, gene therapy for SCI has shown limited success to date due in part to the poor utility of conventional gene vectors.

Olfactory ensheathing cells seeded decellularized scaffold promotes axonal regeneration in spinal cord injury rats.

Spinal cord decellularized (DC) scaffolds can promote axonal regeneration and restore hindlimb motor function of spinal cord defect rats. However, scarring caused by damage to the astrocytes at the margin of injury can hinder axon regeneration. Olfactory ensheathing cells (OECs) integrate and migrate with astrocytes at the site of spinal cord injury, providing a bridge for axons to penetrate the scars and grow into lesion cores.

Unsupervised Person Re-identification via Cross-camera Similarity Exploration.

Most person re-identification (re-ID) approaches are based on supervised learning, which requires manually annotated data. However, it is not only resource-intensive to acquire identity annotation but also impractical for large-scale data. To relieve this problem, we propose a cross-camera unsupervised approach that makes use of unsupervised style-transferred images to jointly optimize a convolutional neural network (CNN) and the relationship among the individual samples for person re-ID.

Antibacterial harziane diterpenoids from a fungal symbiont Trichoderma atroviride isolated from Colquhounia coccinea var. mollis.

Fungal endophytes from plants are an important source for discovery of novel bioactive natural products. In this study, five undescribed harziane diterpenoids with a 4/7/5/6 tetracyclic scaffold, harzianols F‒J and three known derivatives, were obtained from the liquid fermentation of an endophytic fungus Trichoderma atroviride B7, which was isolated from the healthy flower of a Lamiaceae plant Colquhounia coccinea var. mollis.

Progressive Learning for Person Re-Identification with One Example.

In this paper, we focus on the one-example person re-identification (re-ID) task, where each identity has only one labeled example along with many unlabeled examples. We propose a progressive framework which gradually exploits the unlabeled data for person re-ID. In this framework, we iteratively (1) update the Convolutional Neural Network (CNN) model and (2) estimate pseudo labels for the unlabeled data.