Boosting semi-supervised learning with Contrastive Complementary Labeling.

Semi-supervised learning (SSL) approaches have achieved great success in leveraging a large amount of unlabeled data to learn deep models. Among them, one popular approach is pseudo-labeling which generates pseudo labels only for those unlabeled data with high-confidence predictions. As for the low-confidence ones, existing methods often simply discard them because these unreliable pseudo labels may mislead the model. Unlike existing methods, we highlight that these low-confidence data can be still beneficial to the training process. Specifically, although we cannot determine which class a low-confidence sample belongs to, we can assume that this sample should be very unlikely to belong to those classes with the lowest probabilities (often called complementary classes/labels). Inspired by this, we propose a novel Contrastive Complementary Labeling (CCL) method that constructs a large number of reliable negative pairs based on the complementary labels and adopts contrastive learning to make use of all the unlabeled data. Extensive experiments demonstrate that CCL significantly improves the performance on top of existing advanced methods and is particularly effective under the label-scarce settings. For example, CCL yields an improvement of 2.43% over FixMatch on CIFAR-10 only with 40 labeled data.

ISX-9 Promotes KGF Secretion From MSCs to Alleviate ALI Through NGFR-ERK-TAU-ß-Catenin Signaling Axis.

BACKGROUND: Mesenchymal stem cells (MSCs) have been widely studied to alleviate acute lung injury (ALI) due to their paracrine function. However, the microenvironment of inflammatory outbreaks significantly restricted the factors secreted from MSCs like keratinocyte growth factor (KGF). KGF is a growth factor with tissue-repaired ability. Is there a better therapeutic prospect for MSCs in combination with compounds that promote their paracrine function? Through compound screening, we screened out isoxazole-9 (ISX-9) to promote MSCs derived KGF secretion and investigated the underlying mechanisms of action. METHODS: Compounds that promote KGF secretion were screened by a dual-luciferase reporter gene assay. The TMT isotope labeling quantitative technique was used to detect the differential proteins upon ISX-9 administrated to MSCs. The expressions of NGFR, ERK, TAU, and ß-catenin were detected by Western blot. In the ALI model, we measured the inflammatory changes by HE staining, SOD content detection, RT-qPCR, immunofluorescence, etc. The influence of ISX-9 on the residence time of MSCs transplantation was explored by optical in vivo imaging. RESULTS: We found out that ISX-9 can promote the expression of KGF in MSCs. ISX-9 acted on the membrane receptor protein NGFR, upregulated phosphorylation of downstream signaling proteins ERK and TAU, downregulated phosphorylation of ß-catenin, and accelerated ß-catenin into the nucleus to further increase the expression of KGF. In the ALI model, combined ISX-9 with MSCs treatments upgraded the expression of KGF in the lung, and enhanced the effect of MSCs in reducing inflammation and repairing lung damage compared with MSCs alone. CONCLUSIONS: ISX-9 facilitated the secretion of KGF from MSCs both in vivo and in vitro. The combination of ISX-9 with MSCs enhanced the paracrine function and anti-inflammatory effect of MSCs compared with MSCs applied alone in ALI. ISX-9 played a contributive role in the transplantation of MSCs for the treatment of ALI.

Improving fine-tuning of self-supervised models with Contrastive Initialization.

Self-supervised learning (SSL) has achieved remarkable performance in pre-training the models that can be further used in downstream tasks via fine-tuning. However, these self-supervised models may not capture meaningful semantic information since the images belonging to the same class are often regarded as negative pairs in the contrastive loss. Consequently, the images of the same class are often located far away from each other in the learned feature space, which would inevitably hamper the fine-tuning process. To address this issue, we seek to explicitly enhance the semantic relation among instances on the targeted downstream task and provide a better initialization for the subsequent fine-tuning. To this end, we propose a Contrastive Initialization (COIN) method that breaks the standard fine-tuning pipeline by introducing an extra class-aware initialization stage before fine-tuning. Specifically, we exploit a supervised contrastive loss to increase inter-class discrepancy and intra-class compactness of features on the target dataset. In this way, self-supervised models can be easily trained to discriminate instances of different classes during the final fine-tuning stage. Extensive experiments show that, with the enriched semantics, our COIN significantly outperforms existing methods without introducing extra training cost and sets new state-of-the-arts on multiple downstream tasks. For example, compared with the baseline method, our COIN improves the accuracy by 5% on ImageNet-20 and 2.57% on CIFAR100, respectively.

Near-Infrared Fluorescent Probes for the Detection of Cancer-Associated Proteases.

Proteases are enzymes capable of catalyzing protein breakdown, which is critical across many biological processes. There are several families of proteases, each of which perform key functions through the degradation of specific proteins. As our understanding of cancer improves, it has been demonstrated that several proteases can be overactivated during the progression of cancer and contribute to malignancy. Optical imaging systems that employ near-infrared (NIR) fluorescent probes to detect protease activity offer clinical promise, both for early detection of cancer as well as for the assessment of personalized therapy. In this Review, we review the design of NIR probes and their successful application for the detection of different cancer-associated proteases.