Muscular adaptations in swimming scale worms (Polynoidae, Annelida)

Annelids are predominantly found along with the seafloor, but over time have colonized a vast diversity of habitats, such as the water column, where different modes of locomotion are necessary. Yet, little is known about their potential muscular adaptation to the continuous swimming behaviour required in the water column. The musculature and motility were examined for five scale worm species of Polynoidae (Aphroditiformia, Annelida) found in shallow waters, deep sea or caves and which exhibit crawling, occasional swimming or continuous swimming, respectively. Their parapodial musculature was reconstructed using microCT and computational three-dimensional analyses, and the muscular functions were interpreted from video recordings of their locomotion. Since most benthic scale worms are able to swim for short distances using body and parapodial muscle movements, suitable musculature for swimming is already present. Our results indicate that rather than rearrangements or addition of muscles, a shift to a pelagic lifestyle is mainly accompanied by structural loss of muscle bundles and density, as well as elongation of extrinsic dorsal and ventral parapodial muscles. Our study documents clear differences in locomotion and musculature among closely related annelids with different lifestyles as well as points to myoanatomical adaptations for accessing the water column.

Graph Deformer Network

Convolution learning on graphs draws increasing attention recently due to its potential applications to a large amount of irregular data. Most graph convolution methods leverage the plain summation/average aggregation to avoid the discrepancy of responses from isomorphic graphs. However, such an extreme collapsing way would result in a structural loss and signal entanglement of nodes, which further cause the degradation of the learning ability. In this paper, we propose a simple yet effective Graph Deformer Network (GDN) to fulfill anisotropic convolution filtering on graphs, analogous to the standard convolution operation on images. Local neighborhood subgraphs (acting like receptive fields) with different structures are deformed into a unified virtual space, coordinated by several anchor nodes. In the deformation process, we transfer components of nodes therein into affinitive anchors by learning their correlations, and build a multi-granularity feature space calibrated with anchors. Anisotropic convolutional kernels can be further performed over the anchor-coordinated space to well encode local variations of receptive fields. By parameterizing anchors and stacking coarsening layers, we build a graph deformer network in an end-to-end fashion. Theoretical analysis indicates its connection to previous work and shows the promising property of graph isomorphism testing. Extensive experiments on widely-used datasets validate the effectiveness of GDN in graph and node classifications.

Preparation and Evaluation of Virgin Olive Oil Oleogels Including Thyme and Cumin Spices with Sunflower Wax

This study aimed to prepare and evaluate virgin olive oil (VOO) oleogels enriched with thyme and cumin spices with sunflower wax (SW) organogelator. Common physico-chemical, structural, thermal, and rheological analyses were completed. Furthermore, aromatic volatiles composition, sensory descriptive analysis, and consumer tests were provided. Results indicated that spice addition does not interfere with gel formation, stability, and gelation time. The oleogels’ color values were affected by the color of the VOO and the spices. The free fatty acidity and peroxide values were within the acceptable limits for virgin olive oils. There were β’ crystal polymorphs, and melting peak temperatures were around 62 °C. Rheological analyses proved that the oleogels were fairly stable under moderate frequencies, maintained their gelled state until around 52 °C, and recovered their shear induced structural loss after force cessation. There were 22 aromatic volatiles quantified in the samples, which originated from the VOO and spices used as ingredients. A trained panel defined the samples using 13 sensory descriptors. Consumer tests proved that the new oleogels were liked by consumers. Overall, this study provided information and the possibility of spice-enriched and spreadable VOO oleogels to enhance per capita consumption of olive oils with new consumption habits.

Role of wetlands in reducing structural loss is highly dependent on characteristics of storms and local wetland and structure conditions

Coastal communities in New Jersey (NJ), New York (NY), and Connecticut (CT) sustained huge structural loss during Sandy in 2012. We present a comprehensive science-based study to assess the role of coastal wetlands in buffering surge and wave in the tri-state by considering Sandy, a hypothetical Black Swan (BS) storm, and the 1% annual chance flood and wave event. Model simulations were conducted with and without existing coastal wetlands, using a dynamically coupled surge-wave model with two types of coastal wetlands. Simulated surge and wave for Sandy were verified with data at numerous stations. Structural loss estimated using real property data and latest damage functions agreed well with loss payout data. Results show that, on zip-code scale, the relative structural loss varies significantly with the percent wetland cover, the at-risk structural value, and the average wave crest height. Reduction in structural loss by coastal wetlands was low in Sandy, modest in the BS storm, and significant in the 1% annual chance flood and wave event. NJ wetlands helped to avoid 8%, 26%, 52% loss during Sandy, BS storm, and 1% event, respectively. This regression model can be used for wetland restoration planning to further reduce structural loss in coastal communities.

An improved efficient model for structure-aware lane detection of unmanned vehicles

Lane detection algorithms require extremely low computational costs as an important part of autonomous driving. Due to heavy backbone networks, algorithms based on pixel-wise segmentation is struggling to handle the problem of runtime consumption in the recognition of lanes. In this paper, a novel and practical methodology based on lightweight Segmentation Network is proposed, which aims to achieve accurate and efficient lane detection. Different with traditional convolutional layers, the proposed Shadow module can reduce the computational cost of the backbone network by performing linear transformations on intrinsic feature maps. Thus a lightweight backbone network Shadow-VGG-16 is built. After that, a tailored pyramid parsing module is introduced to collect different sub-domain features, which is composed of both a strip pool module based on Pyramid Scene Parsing Network (PSPNet) and a convolution attention module. Finally, a lane structural loss is proposed to explicitly model the lane structure and reduce the influence of noise, so that the pixels can fit the lane better. Extensive experimental results demonstrate that the performance of our method is significantly better than the state-of-the-art (SOTA) algorithms such as Pointlanenet and Line-CNN et al. 95.28% and 90.06% accuracy and 62.5 frames per second (fps) inference speed can be achieved on the CULane and Tusimple test dataset. Compared with the latest ERFNet, Line-CNN, SAD, F1 scores have respectively increased by 3.51%, 2.84%, and 3.82%. Meanwhile, the result from our dataset exceeds the top performances of the other by 8.6% with an 87.09 F1 score, which demonstrates the superiority of our method.

A Generative Adversarial Network-Based Image Denoiser Controlling Heterogeneous Losses

We propose a novel generative adversarial network (GAN)-based image denoising method that utilizes heterogeneous losses. In order to improve the restoration quality of the structural information of the generator, the heterogeneous losses, including the structural loss in addition to the conventional mean squared error (MSE)-based loss, are used to train the generator. To maximize the improvements brought on by the heterogeneous losses, the strength of the structural loss is adaptively adjusted by the discriminator for each input patch. In addition, a depth wise separable convolution-based module that utilizes the dilated convolution and symmetric skip connection is used for the proposed GAN so as to reduce the computational complexity while providing improved denoising quality compared to the convolutional neural network (CNN) denoiser. The experiments showed that the proposed method improved visual information fidelity and feature similarity index values by up to 0.027 and 0.008, respectively, compared to the existing CNN denoiser.

CliGAN: A Structurally Sensitive Convolutional Neural Network Model for Statistical Downscaling of Precipitation from Multi-Model Ensembles

Despite numerous studies in statistical downscaling methodologies, there remains a lack of methods that can downscale from precipitation modeled in global climate models to regional level high resolution gridded precipitation. This paper reports a novel downscaling method using a Generative Adversarial Network (GAN), CliGAN, which can downscale large-scale annual maximum precipitation given by simulation of multiple atmosphere-ocean global climate models (AOGCM) from Coupled Model Inter-comparison Project 6 (CMIP6) to regional-level gridded annual maximum precipitation data. This framework utilizes a convolution encoder-dense decoder network to create a generative network and a similar network to create a critic network. The model is trained using an adversarial training approach. The critic uses the Wasserstein distance loss function and the generator is trained using a combination of adversarial loss Wasserstein distance, structural loss with the multi-scale structural similarity index (MSSIM), and content loss with the Nash-Sutcliff Model Efficiency (NS). The MSSIM index allowed us to gain insight into the model’s regional characteristics and shows that relying exclusively on point-based error functions, widely used in statistical downscaling, may not be enough to reliably simulate regional precipitation characteristics. Further use of structural loss functions within CNN-based downscaling methods may lead to higher quality downscaled climate model products.

CliGAN: A Structurally Sensitive Convolutional Neural Network Model for Statistical Downscaling of Precipitation from Multi-Model Ensembles

Despite numerous studies in statistical downscaling methodology, there remains a lack of methods that can downscale from precipitation modeled in global climate models to regional level high resolution gridded precipitation. This paper reports a novel downscaling method using a Generative Adversarial Network (GAN), CliGAN, which can downscale large-scale annual maximum precipitation given by simulation of multiple atmosphere-ocean global climate models (AOGCM) from Coupled Model Inter-comparison Project 6 (CMIP6) to regional-level gridded annual maximum precipitation data. This framework utilizes a convolution encoder-dense decoder network to create a generative network and a similar network to create a critic network. The model is trained using an adversarial training approach. The critic uses the Wasserstein distance loss function and the generator is trained using a combination of adversarial loss Wasserstein distance, structural loss with the multi-scale structural similarity index (MSSIM), and content loss with the Nash-Sutcliff Model Efficiency (NS). The MSSIM index allowed us to gain insight into the model’s regional characteristics and shows that relying exclusively on point-based error functions, widely used in statistical downscaling, may not be enough to reliably simulate regional precipitation characteristics. Further use of structural loss functions within CNN-based downscaling methods may lead to higher quality downscaled climate model products.

Melatonin alleviates pyroptosis of retinal neurons following acute intraocular hypertension

Background: Glaucoma is a multifactorial optic neuropathy progressive characterized by structural loss of retinal ganglion cells (RGCs) and irreversible loss of vision. High intraocular pressure (HIOP) is a high risk factor for glaucoma. It has been reported that the manners of RGCs’ loss are in-depth explored after acute HIOP injury, such as, apoptosis, autophagy and necrosis. However, pyroptosis, a novel type of pro-inflammatory cell programmed necrosis, rarely reported after acute HIOP injury. Researches also showed that melatonin (MT) possesses substantial anti-inflammatory properties. However, whether melatonin could alleviate retinal neurons death, especially pyroptosis, by acute HIOP injury is unclear. Objective: This study explored pyroptosis of retinal neurons and the effects of MT preventing retinal neurons form pyroptosis after acute HIOP injury. Method: Establish acute HIOP model in rat by increasing the IOP and then reperfusion. Western Blot (WB) was adopted to detect molecules related to pyroptosis at the protein level, such as GasderminD (GSDMD), GasderminDp32 (GSDMDp32), Caspase-1 (Casp-1) and Caspase-1p20 (Casp-1p20), and the products of inflammatory reactions, as interleukin-18 (IL-18) and interleukin-1β (IL-1β) as well. At the same time, Immunofluorescence (IF) was used to co-localize Casp-1with retinal neurons to determine the position of Casp-1 expression. Morphologically, Ethidium homodimer-III staining, a method commonly used for judging cell death, was carried out to stain dead cells. Subsequently, Lactate Dehydrogenase (LDH) cytotoxicity assay kit was used to quantitative analysis the LDH released after cell disruption. Results: The results suggested that pyroptosis played a vital role in retinal neurons death, especially in the ganglion cell layer, by acute HIOP injury and peaked at 6h after acute HIOP injury. Furthermore, it was found that MT might prevent retinal neurons from pyroptosis via NF-κB/NLRP3 axis after acute HIOP injury in rats. Conclusion: MT treatment might be considered a new strategy for protecting retinal neurons against pyroptosis following acute HIOP injury.