Transformers have recently gained significant attention in the computer vision community. However, the lack of scalability of self-attention mechanisms with respect to image size has limited their wide adoption in state-of-the-art vision backbones. In this paper we introduce an efficient and scalable attention model we call multi-axis attention, which consists of two aspects: blocked local and dilated global attention. These design choices allow global-local spatial interactions on arbitrary input resolutions with only linear complexity. We also present a new architectural element by effectively blending our proposed attention model with convolutions, and accordingly propose a simple hierarchical vision backbone, dubbed MaxViT, by simply repeating the basic building block over multiple stages. Notably, MaxViT is able to ''see'' globally throughout the entire network, even in earlier, high-resolution stages. We demonstrate the effectiveness of our model on a broad spectrum of vision tasks. On image classification, MaxViT achieves state-of-the-art performance under various settings: without extra data, MaxViT attains 86.5% ImageNet-1K top-1 accuracy; with ImageNet-21K pre-training, our model achieves 88.7% top-1 accuracy. For downstream tasks, MaxViT as a backbone delivers favorable performance on object detection as well as visual aesthetic assessment. We also show that our proposed model expresses strong generative modeling capability on ImageNet, demonstrating the superior potential of MaxViT blocks as a universal vision module. The source code and trained models will be available at https://github.com/google-research/maxvit.

MaxViT: Multi-Axis Vision Transformer

Regulated cell death (RCD), also well-known as programmed cell death (PCD), refers to the form of cell death that can be regulated by a variety of biomacromolecules, which is distinctive from accidental cell death (ACD). Accumulating evidence has revealed that RCD subroutines are the key features of tumorigenesis, which may ultimately lead to the establishment of different potential therapeutic strategies. Hitherto, targeting the subroutines of RCD with pharmacological small-molecule compounds has been emerging as a promising therapeutic avenue, which has rapidly progressed in many types of human cancers. Thus, in this review, we focus on summarizing not only the key apoptotic and autophagy-dependent cell death signaling pathways, but the crucial pathways of other RCD subroutines, including necroptosis, pyroptosis, ferroptosis, parthanatos, entosis, NETosis and lysosome-dependent cell death (LCD) in cancer. Moreover, we further discuss the current situation of several small-molecule compounds targeting the different RCD subroutines to improve cancer treatment, such as single-target, dual or multiple-target small-molecule compounds, drug combinations, and some new emerging therapeutic strategies that would together shed new light on future directions to attack cancer cell vulnerabilities with small-molecule drugs targeting RCD for therapeutic purposes. 

/pdf/regulated-cell-death-rcd-in-cancer-key-pathways-and-targeted-1zhjnd92.pdf

Regulated cell death (RCD) in cancer: key pathways and targeted therapies

Vision transformer (ViT) recently has drawn great attention in computer vision due to its remarkable model capability. However, most prevailing ViT models suffer from huge number of parameters, restricting their applicability on devices with limited resources. To alleviate this issue, we propose TinyViT, a new family of tiny and efficient small vision transformers pretrained on large-scale datasets with our proposed fast distillation framework. The central idea is to transfer knowledge from large pretrained models to small ones, while enabling small models to get the dividends of massive pretraining data. More specifically, we apply distillation during pretraining for knowledge transfer. The logits of large teacher models are sparsified and stored in disk in advance to save the memory cost and computation overheads. The tiny student transformers are automatically scaled down from a large pretrained model with computation and parameter constraints. Comprehensive experiments demonstrate the efficacy of TinyViT. It achieves a top-1 accuracy of 84.8% on ImageNet-1k with only 21M parameters, being comparable to Swin-B pretrained on ImageNet-21k while using 4.2 times fewer parameters. Moreover, increasing image resolutions, TinyViT can reach 86.5% accuracy, being slightly better than Swin-L while using only 11% parameters. Last but not the least, we demonstrate a good transfer ability of TinyViT on various downstream tasks. Code and models are available at https://github.com/microsoft/Cream/tree/main/TinyViT.

TinyViT: Fast Pretraining Distillation for Small Vision Transformers

Vision Transformers convert images to sequences by slicing them into patches. The size of these patches controls a speed/accuracy tradeoff, with smaller patches leading to higher accuracy at greater computational cost, but changing the patch size typically requires retraining the model. In this paper, we demonstrate that simply randomizing the patch size at training time leads to a single set of weights that performs well across a wide range of patch sizes, making it possible to tailor the model to different compute budgets at deployment time. We extensively evaluate the resulting model, which we call FlexiViT, on a wide range of tasks, including classification, image-text retrieval, open-world detection, panoptic segmentation, and semantic segmentation, concluding that it usually matches, and sometimes outperforms, standard ViT models trained at a single patch size in an otherwise identical setup. Hence, FlexiViT training is a simple drop-in improvement for ViT that makes it easy to add compute-adaptive capabilities to most models relying on a ViT backbone architecture. Code and pre-trained models are available at https://github.com/google-research/big_vision

FlexiViT: One Model for All Patch Sizes

Silicone-based polymer composites containing high atomic number additives are prioritized for the development of new materials for radiation shielding, due to their mechanical, thermal, electrical, and multifunctional properties. The X-ray attenuation properties, as well as mechanical properties, of the newly developed-lead-free multi-layered structures for radiation shielding, based on silicone composite layers containing tin, cerium oxide, tungsten oxide, and bismuth additives, are analyzed and discussed in this paper. It is shown that, by varying the additive concentrations in silicone composites, lead-free and flexible layered structures, exhibiting lead-equivalent X-ray shielding, can be fabricated.

/pdf/lead-free-multilayered-polymer-composites-for-radiation-2a1sn4ti.pdf

Lead Free Multilayered Polymer Composites for Radiation Shielding

As a typical problem in time series analysis, traffic flow prediction is one of the most important application fields of machine learning. However, achieving highly accurate traffic flow prediction is a challenging task, due to the presence of complex dynamic spatial-temporal dependencies within a road network. This paper proposes a novel Dynamic Spatial-Temporal Aware Graph Neural Network (DSTAGNN) to model the complex spatial-temporal interaction in road network. First, considering the fact that historical data carries intrinsic dynamic information about the spatial structure of road networks, we propose a new dynamic spatial-temporal aware graph based on a data-driven strategy to replace the pre-defined static graph usually used in traditional graph convolution. Second, we design a novel graph neural network architecture, which can not only represent dynamic spatial relevance among nodes with an improved multi-head attention mechanism, but also acquire the wide range of dynamic temporal dependency from multi-receptive field features via multi-scale gated convolution. Extensive experiments on real-world data sets demonstrate that our proposed method significantly outperforms the state-of-the-art methods.

DSTAGNN: Dynamic Spatial-Temporal Aware Graph Neural Network for Traffic Flow Forecasting

Triple-negative breast cancer (TNBC) is a subtype of human breast cancer with one of the worst prognoses, with no targeted therapeutic strategies currently available. Regulated cell death (RCD), also known as programmed cell death (PCD), has been widely reported to have numerous links to the progression and therapy of many types of human cancer. Of note, RCD can be divided into numerous different subroutines, including autophagy-dependent cell death, apoptosis, mitotic catastrophe, necroptosis, ferroptosis, pyroptosis and anoikis. More recently, targeting the subroutines of RCD with small-molecule compounds has been emerging as a promising therapeutic strategy, which has rapidly progressed in the treatment of TNBC. Therefore, in this review, we focus on summarizing the molecular mechanisms of the above-mentioned seven major RCD subroutines related to TNBC and the latest progress of small-molecule compounds targeting different RCD subroutines. Moreover, we further discuss the combined strategies of one drug (e.g., narciclasine) or more drugs (e.g., torin-1 combined with chloroquine) to achieve the therapeutic potential on TNBC by regulating RCD subroutines. More importantly, we demonstrate several small-molecule compounds (e.g., ONC201 and NCT03733119) by targeting the subroutines of RCD in TNBC clinical trials. Taken together, these findings will provide a clue on illuminating more actionable low-hanging-fruit druggable targets and candidate small-molecule drugs for potential RCD-related TNBC therapies. 

/pdf/targeting-regulated-cell-death-rcd-with-small-molecule-1xweukw3.pdf

Targeting regulated cell death (RCD) with small-molecule compounds in triple-negative breast cancer: a revisited perspective from molecular mechanisms to targeted therapies

This work targets automated designing and scaling of Vision Transformers (ViTs). The motivation comes from two pain spots: 1) the lack of efﬁcient and principled methods for designing and scaling ViTs; 2) the tremendous computational cost of training ViT that is much heavier than its convolution counterpart. To tackle these issues, we propose As-ViT , an auto-scaling framework for ViTs without training, which automatically discovers and scales up ViTs in an efﬁcient and principled manner. Speciﬁcally, we ﬁrst design a “seed” ViT topology by leveraging a training-free search process. This extremely fast search is fulﬁlled by a comprehensive study of ViT’s network complexity, yielding a strong Kendall-tau correlation with ground-truth accuracies. Second, starting from the “seed” topology, we automate the scaling rule for ViTs by growing widths/depths to different ViT layers. This results in a series of architectures with different numbers of parameters in a single run. Finally, based on the observation that ViTs can tolerate coarse tokenization in early training stages, we propose a progressive tokenization strategy to train ViTs faster and cheaper. As a uniﬁed framework, As-ViT achieves strong performance on classiﬁcation (83.5% top1 on ImageNet-1k) and detection (52.7% mAP on COCO) without any manual crafting nor scaling of ViT architectures: the end-to-end model design and scaling process costs only 12 hours on one V100 GPU . Our code is available at https://github.com/VITA-Group/AsViT .

Auto-scaling Vision Transformers without Training

An improved deep neural network incorporating attention mechanism and DSSIM loss function (AM_U_Net) is used to recover input images with speckles transmitted through a multimode fiber (MMF). The network is trained on a relatively small dataset and demonstrates an optimal reconstruction ability and generalization ability. Furthermore, a bimodal fusion method is developed based on S polarization and P polarization speckles, greatly improving the recognition accuracy. These findings prove that AM_U_Net has remarkable capabilities for information recovery and transfer learning and good tolerance and robustness under different MMF transmission conditions, indicating its significant application potential in medical imaging and secure communication.

Deep learning image transmission through a multimode fiber based on a small training dataset.

Homogeneous instance segmentation aims to identify each instance in an image where all interested instances belong to the same category, such as plant leaves and microscopic cells. Recently, proposal-free methods, which straightforwardly generate instance-aware information to group pixels into different instances, have received increasing attention due to their efficient pipeline. However, they often fail to distinguish adjacent instances due to similar appearances, dense distribution and ambiguous boundaries of instances in homogeneous images. In this paper, we propose a pixel-embedded affinity modeling method for homogeneous instance segmentation, which is able to preserve the semantic information of instances and improve the distinguishability of adjacent instances. Instead of predicting affinity directly, we propose a self-correlation module to explicitly model the pairwise relationships between pixels, by estimating the similarity between embeddings generated from the input image through CNNs. Based on the self-correlation module, we further design a cross-correlation module to maintain the semantic consistency between instances. Specifically, we map the transformed input images with different views and appearances into the same embedding space, and then mutually estimate the pairwise relationships of embeddings generated from the original input and its transformed variants. In addition, to integrate the global instance information, we introduce an embedding pyramid module to model affinity on different scales. Extensive experiments demonstrate the versatile and superior performance of our method on three representative datasets. Code and models are available at https://github.com/weih527/Pixel-Embedded-Affinity.

/pdf/learning-to-model-pixel-embedded-affinity-for-homogeneous-2eqd0kdr.pdf

Wei Huang

Papers

DSTAGNN: Dynamic Spatial-Temporal Aware Graph Neural Network for Traffic Flow Forecasting

Targeting regulated cell death (RCD) with small-molecule compounds in triple-negative breast cancer: a revisited perspective from molecular mechanisms to targeted therapies

Auto-scaling Vision Transformers without Training

Deep learning image transmission through a multimode fiber based on a small training dataset.

Learning to Model Pixel-Embedded Affinity for Homogeneous Instance Segmentation