Cancer-associated snaR-A noncoding RNA interacts with core splicing machinery and disrupts processing of mRNA subpopulations.

RNA polymerase III (Pol III) activity in cancer is linked to the production of small noncoding (nc)RNAs that are otherwise silent in most tissues. snaR-A (small NF90-associated RNA isoform A) - a hominid-specific ncRNA shown to enhance cell proliferation, migration, and invasion - is a cancer-emergent Pol III product that remains largely uncharacterized despite promoting growth phenotypes. Here, we applied a combination of genomic and biochemical approaches to study the biogenesis and subsequent protein interactions of snaR-A and to better understand its role as a putative driver of cancer progression. By profiling the chromatin landscapes across a multitude of primary tumor types, we show that predicted snaR-A upregulation is broadly linked with unfavorable outcomes among cancer patients. At the molecular level, we unexpectedly discover widespread interactions between snaR-A and mRNA splicing factors, including SF3B2 - a core component of the U2 small nuclear ribonucleoprotein (snRNP). We find that SF3B2 levels are sensitive to high snaR-A abundance and that depletion of snaR-A alone is sufficient to decrease intron retention levels across subpopulations of mRNA enriched for U2 snRNP occupancy. snaR-A sensitive genes are characterized by high GC content, close spatial proximity to nuclear bodies concentrated in pre-mRNA splicing factors, and functional enrichment for proteins involved in deacetylation and autophagy. We highlight examples of splicing misregulation and increased protein levels following snaR-A depletion for a wide-ranging set of factors, suggesting snaR-A-driven splicing defects may have far-reaching effects that re-shape the cellular proteome. These findings clarify the molecular activities and consequences of snaR-A in cancer, and altogether establish a novel mechanism through which Pol III overactivity may promote tumorigenesis.

A Survey of Knowledge Graph Reasoning on Graph Types: Static, Dynamic, and Multi-Modal.

Knowledge graph reasoning (KGR), aiming to deduce new facts from existing facts based on mined logic rules underlying knowledge graphs (KGs), has become a fast-growing research direction. It has been proven to significantly benefit the usage of KGs in many AI applications, such as question answering, recommendation systems, and etc. According to the graph types, existing KGR models can be roughly divided into three categories, i.e., static models, temporal models, and multi-modal models. Early works in this domain mainly focus on static KGR, and recent works try to leverage the temporal and multi-modal information, which are more practical and closer to real-world. However, no survey papers and open-source repositories comprehensively summarize and discuss models in this important direction. To fill the gap, we conduct a first survey for knowledge graph reasoning tracing from static to temporal and then to multi-modal KGs. Concretely, the models are reviewed based on bi-level taxonomy, i.e., top-level (graph types) and base-level (techniques and scenarios). Besides, the performances, as well as datasets, are summarized and presented. Moreover, we point out the challenges and potential opportunities to enlighten the readers. The corresponding open-source repository is shared on GitHub https://github.com/LIANGKE23/Awesome-Knowledge-Graph-Reasoning.

Evidence of RNA polymerase III recruitment and transcription at protein-coding gene promoters.

RNA polymerase (Pol) I, II, and III are most commonly described as having distinct roles in synthesizing ribosomal RNA (rRNA), messenger RNA (mRNA), and specific small noncoding (nc)RNAs, respectively. This delineation of transcriptional responsibilities is not definitive, however, as evidenced by instances of Pol II recruitment to genes conventionally transcribed by Pol III, including the co-transcription of RPPH1 - the catalytic RNA component of RNase P. A comprehensive understanding of the interplay between RNA polymerase complexes remains lacking, however, due to limited comparative analyses for all three enzymes. To address this gap, we applied a uniform framework for quantifying global Pol I, II, and III occupancies that integrates currently available human RNA polymerase ChIP-seq datasets. Occupancy maps are combined with a comprehensive multi-class promoter set that includes protein-coding genes, noncoding genes, and repetitive elements. While our genomic survey appropriately identifies recruitment of Pol I, II, and III to canonical target genes, we unexpectedly discover widespread recruitment of the Pol III machinery to promoters of specific protein-coding genes, supported by colocalization patterns observed for several Pol III-specific subunits. We show that Pol III-occupied Pol II promoters are enriched for small, nascent RNA reads terminating in a run of 4 Ts, a unique hallmark of Pol III transcription termination and evidence of active Pol III activity at these sites. Pol III disruption differentially modulates the expression of Pol III-occupied coding genes, which are functionally enriched for ribosomal proteins and genes broadly linked to unfavorable outcomes in cancer. Our map also identifies additional, currently unannotated genomic elements occupied by Pol III with clear signatures of nascent RNA species that are sensitive to disruption of La (SSB) - a Pol III-related RNA chaperone protein. These findings reshape our current understanding of the interplay between Pols II and III and identify potentially novel small ncRNAs with broad implications for gene regulatory paradigms and RNA biology.

Self-Supervised Temporal Graph Learning With Temporal and Structural Intensity Alignment.

Temporal graph learning aims to generate high-quality representations for graph-based tasks with dynamic information, which has recently garnered increasing attention. In contrast to static graphs, temporal graphs are typically organized as node interaction sequences over continuous time rather than an adjacency matrix. Most temporal graph learning methods model current interactions by incorporating historical neighborhood. However, such methods only consider first-order temporal information while disregarding crucial high-order structural information, resulting in suboptimal performance. To address this issue, we propose a self-supervised method called S2T for temporal graph learning, which extracts both temporal and structural information to learn more informative node representations. Notably, the initial node representations combine first-order temporal and high-order structural information differently to calculate two conditional intensities. An alignment loss is then introduced to optimize the node representations, narrowing the gap between the two intensities and making them more informative. Concretely, in addition to modeling temporal information using historical neighbor sequences, we further consider structural knowledge at both local and global levels. At the local level, we generate structural intensity by aggregating features from high-order neighbor sequences. At the global level, a global representation is generated based on all nodes to adjust the structural intensity according to the active statuses on different nodes. Extensive experiments demonstrate that the proposed model S2T achieves at most 10.13% performance improvement compared with the state-of-the-art competitors on several datasets.

SARF: Aliasing Relation-Assisted Self-Supervised Learning for Few-Shot Relation Reasoning.

Few-shot relation reasoning on knowledge graphs (FS-KGR) is an important and practical problem that aims to infer long-tail relations and has drawn increasing attention these years. Among all the proposed methods, self-supervised learning (SSL) methods, which effectively extract the hidden essential inductive patterns relying only on the support sets, have achieved promising performance. However, the existing SSL methods simply cut down connections between high-frequency and long-tail relations, which ignores the fact, i.e., the two kinds of information could be highly related to each other. Specifically, we observe that relations with similar contextual meanings, called aliasing relations (ARs), may have similar attributes. In other words, the ARs of the target long-tail relation could be in high-frequency, and leveraging such attributes can largely improve the reasoning performance. Based on the interesting observation above, we proposed a novel Self-supervised learning model by leveraging Aliasing Relations to assist FS-KGR, termed . Specifically, we propose a graph neural network (GNN)-based AR-assist module to encode the ARs. Besides, we further provide two fusion strategies, i.e., simple summation and learnable fusion, to fuse the generated representations, which contain extra abundant information underlying the ARs, into the self-supervised reasoning backbone for performance enhancement. Extensive experiments on three few-shot benchmarks demonstrate that achieves state-of-the-art (SOTA) performance compared with other methods in most cases.

Revisiting Initializing Then Refining: An Incomplete and Missing Graph Imputation Network.

With the development of various applications, such as recommendation systems and social network analysis, graph data have been ubiquitous in the real world. However, graphs usually suffer from being absent during data collection due to copyright restrictions or privacy-protecting policies. The graph absence could be roughly grouped into attribute-incomplete and attribute-missing cases. Specifically, attribute-incomplete indicates that a portion of the attribute vectors of all nodes are incomplete, while attribute-missing indicates that all attribute vectors of partial nodes are missing. Although various graph imputation methods have been proposed, none of them is custom-designed for a common situation where both types of graph absence exist simultaneously. To fill this gap, we develop a novel graph imputation network termed revisiting initializing then refining (RITR), where both attribute-incomplete and attribute-missing samples are completed under the guidance of a novel initializing-then-refining imputation criterion. Specifically, to complete attribute-incomplete samples, we first initialize the incomplete attributes using Gaussian noise before network learning, and then introduce a structure-attribute consistency constraint to refine incomplete values by approximating a structure-attribute correlation matrix to a high-order structure matrix. To complete attribute-missing samples, we first adopt structure embeddings of attribute-missing samples as the embedding initialization, and then refine these initial values by adaptively aggregating the reliable information of attribute-incomplete samples according to a dynamic affinity structure. To the best of our knowledge, this newly designed method is the first end-to-end unsupervised framework dedicated to handling hybrid-absent graphs. Extensive experiments on six datasets have verified that our methods consistently outperform the existing state-of-the-art competitors. Our source code is available at https://github.com/WxTu/RITR.

A Combinatorial Regulatory Platform Determines Expression of RNA Polymerase III Subunit RPC7α (POLR3G) in Cancer.

RNA polymerase III (Pol III) subunit RPC7α, which is encoded by POLR3G in humans, has been linked to both tumor growth and metastasis. Accordantly, high POLR3G expression is a negative prognostic factor in multiple cancer subtypes. To date, the mechanisms underlying POLR3G upregulation have remained poorly defined. We performed a large-scale genomic survey of mRNA and chromatin signatures to predict drivers of POLR3G expression in cancer. Our survey uncovers positive determinants of POLR3G expression, including a gene-internal super-enhancer bound with multiple transcription factors (TFs) that promote POLR3G expression, as well as negative determinants that include gene-internal DNA methylation, retinoic-acid induced differentiation, and MXD4-mediated disruption of POLR3G expression. We show that novel TFs identified in our survey, including ZNF131 and ZNF207, functionally enhance POLR3G expression, whereas MXD4 likely obstructs MYC-driven expression of POLR3G and other growth-related genes. Integration of chromatin architecture and gene regulatory signatures identifies additional factors, including histone demethylase KDM5B, as likely influencers of POLR3G gene activity. Taken together, our findings support a model in which POLR3G expression is determined with multiple factors and dynamic regulatory programs, expanding our understanding of the circuitry underlying POLR3G upregulation and downstream consequences in cancer.

Hierarchically Contrastive Hard Sample Mining for Graph Self-Supervised Pretraining.

Contrastive learning has recently emerged as a powerful technique for graph self-supervised pretraining (GSP). By maximizing the mutual information (MI) between a positive sample pair, the network is forced to extract discriminative information from graphs to generate high-quality sample representations. However, we observe that, in the process of MI maximization (Infomax), the existing contrastive GSP algorithms suffer from at least one of the following problems: 1) treat all samples equally during optimization and 2) fall into a single contrasting pattern within the graph. Consequently, the vast number of well-categorized samples overwhelms the representation learning process, and limited information is accumulated, thus deteriorating the learning capability of the network. To solve these issues, in this article, by fusing the information from different views and conducting hard sample mining in a hierarchically contrastive manner, we propose a novel GSP algorithm called hierarchically contrastive hard sample mining (HCHSM). The hierarchical property of this algorithm is manifested in two aspects. First, according to the results of multilevel MI estimation in different views, the MI-based hard sample selection (MHSS) module keeps filtering the easy nodes and drives the network to focus more on hard nodes. Second, to collect more comprehensive information for hard sample learning, we introduce a hierarchically contrastive scheme to sequentially force the learned node representations to involve multilevel intrinsic graph features. In this way, as the contrastive granularity goes finer, the complementary information from different levels can be uniformly encoded to boost the discrimination of hard samples and enhance the quality of the learned graph embedding. Extensive experiments on seven benchmark datasets indicate that the HCHSM performs better than other competitors on node classification and node clustering tasks. The source code of HCHSM is available at https://github.com/WxTu/HCHSM.

The preoperative platelet to neutrophil ratio and lymphocyte to monocyte ratio are superior prognostic indicators compared with other inflammatory biomarkers in ovarian cancer.

Background: Previous studies have suggested that the ratios of immune-inflammatory cells could serve as prognostic indicators in ovarian cancer. However, which of these is the superior prognostic indicator in ovarian cancer remains unknown. In addition, studies on the prognostic value of the platelet to neutrophil ratio (PNR) in ovarian cancer are still limited. Methods: A cohort of 991 ovarian cancer patients was analyzed in the present study. Receiver operator characteristic (ROC) curves were utilized to choose the optimal cut-off values of inflammatory biomarkers such as neutrophil to lymphocyte ratio (NLR), lymphocyte to monocyte ratio (LMR), platelet to lymphocyte ratio (PLR), systemic immune-inflammation index (SII), and PNR. The correlation of inflammatory biomarkers with overall survival (OS) and relapse-free survival (RFS) was investigated by Kaplan-Meier methods and log-rank test, followed by Cox regression analyses. Results: Kaplan-Meier curves suggested that LMR<3.39, PLR≥181.46, and PNR≥49.20 had obvious associations with worse RFS (P<0.001, P=0.018, P<0.001). Multivariate analysis suggested that LMR (≥3.39 vs. <3.39) (P=0.042, HR=0.810, 95% CI=0.661-0.992) and PNR (≥49.20 vs. <49.20) (P=0.004, HR=1.351, 95% CI=1.103-1.656) were independent prognostic indicators of poor RFS. In addition, Kaplan-Meier curves indicated that PLR≥182.23 was significantly correlated with worse OS (P=0.039). Conclusion: Taken together, PNR and LMR are superior prognostic indicators compared with NLR, PLR, and SII in patients with ovarian cancer.

scDFC: A deep fusion clustering method for single-cell RNA-seq data.

Clustering methods have been widely used in single-cell RNA-seq data for investigating tumor heterogeneity. Since traditional clustering methods fail to capture the high-dimension methods, deep clustering methods have drawn increasing attention these years due to their promising strengths on the task. However, existing methods consider either the attribute information of each cell or the structure information between different cells. In other words, they cannot sufficiently make use of all of this information simultaneously. To this end, we propose a novel single-cell deep fusion clustering model, which contains two modules, i.e. an attributed feature clustering module and a structure-attention feature clustering module. More concretely, two elegantly designed autoencoders are built to handle both features regardless of their data types. Experiments have demonstrated the validity of the proposed approach, showing that it is efficient to fuse attributes, structure, and attention information on single-cell RNA-seq data. This work will be further beneficial for investigating cell subpopulations and tumor microenvironment. The Python implementation of our work is now freely available at https://github.com/DayuHuu/scDFC.

Simple Contrastive Graph Clustering.

Contrastive learning has recently attracted plenty of attention in deep graph clustering due to its promising performance. However, complicated data augmentations and time-consuming graph convolutional operations undermine the efficiency of these methods. To solve this problem, we propose a simple contrastive graph clustering (SCGC) algorithm to improve the existing methods from the perspectives of network architecture, data augmentation, and objective function. As to the architecture, our network includes two main parts, that is, preprocessing and network backbone. A simple low-pass denoising operation conducts neighbor information aggregation as an independent preprocessing, and only two multilayer perceptrons (MLPs) are included as the backbone. For data augmentation, instead of introducing complex operations over graphs, we construct two augmented views of the same vertex by designing parameter unshared Siamese encoders and perturbing the node embeddings directly. Finally, as to the objective function, to further improve the clustering performance, a novel cross-view structural consistency objective function is designed to enhance the discriminative capability of the learned network. Extensive experimental results on seven benchmark datasets validate our proposed algorithm's effectiveness and superiority. Significantly, our algorithm outperforms the recent contrastive deep clustering competitors with at least seven times speedup on average. The code of SCGC is released at SCGC. Besides, we share a collection of deep graph clustering, including papers, codes, and datasets at ADGC.

Task-Related Saliency for Few-Shot Image Classification.

A weakness of the existing metric-based few-shot classification method is that task-unrelated objects or backgrounds may mislead the model since the small number of samples in the support set is insufficient to reveal the task-related targets. An essential cue of human wisdom in the few-shot classification task is that they can recognize the task-related targets by a glimpse of support images without being distracted by task-unrelated things. Thus, we propose to explicitly learn task-related saliency features and make use of them in the metric-based few-shot learning schema. We divide the tackling of the task into three phases, namely, the modeling, the analyzing, and the matching. In the modeling phase, we introduce a saliency sensitive module (SSM), which is an inexact supervision task jointly trained with a standard multiclass classification task. SSM not only enhances the fine-grained representation of feature embedding but also can locate the task-related saliency features. Meanwhile, we propose a self-training-based task-related saliency network (TRSN) which is a lightweight network to distill task-related salience produced by SSM. In the analyzing phase, we freeze TRSN and use it to handle novel tasks. TRSN extracts task-relevant features while suppressing the disturbing task-unrelated features. We, therefore, can discriminate samples accurately in the matching phase by strengthening the task-related features. We conduct extensive experiments on five-way 1-shot and 5-shot settings to evaluate the proposed method. Results show that our method achieves a consistent performance gain on benchmarks and achieves the state-of-the-art.

Prognostic Value of Liver Kinase B1 (LKB1) in Gastric Cancer-Associated Tumor Microenvironment Immunity.

Liver kinase B1 (LKB1) is a tumor suppressor gene, the inactivation of which occurs frequently in different tumor types. However, whether LKB1 is associated with the clinical features of gastric cancer (GC) and regulating tumor immunity is unknown. In this study, we showed that LKB1 is highly expressed in the serum of healthy individuals (n = 176) compared to GC patients (n = 416) and is also associated with clinical outcomes and good survival rates in GC patients. Furthermore, genes associated with immune checkpoints and T cell activation, such as PD-1, PD-L1, CD8A, CD8B, CD28, and GZMM, were shown to be highly expressed in GC subgroups with high LKB1 expression. Compared with fresh gastric cancerous tissues, LKB1 was highly expressed in CD3+CD8+ and CD3+CD8+CD28+ T cells in fresh adjacent non-cancerous tissues. CD3+CD8+ T cells produced an IFN-γ anti-cancer immune response. Furthermore, the proportion of CD3+CD8+ T cells that expressed LKB had a positive correlation with IFN-γ expression. Moreover, GC patients with low LKB1 expression had a poor objective response rate, and worse progression-free survival and overall survival when treated with pembrolizumab. In conclusion, LKB1 may be a potential immune checkpoint in GC patients.

The Pol III transcriptome: Basic features, recurrent patterns, and emerging roles in cancer.

The RNA polymerase III (Pol III) transcriptome is universally comprised of short, highly structured noncoding RNA (ncRNA). Through RNA-protein interactions, the Pol III transcriptome actuates functional activities ranging from nuclear gene regulation (7SK), splicing (U6, U6atac), and RNA maturation and stability (RMRP, RPPH1, Y RNA), to cytoplasmic protein targeting (7SL) and translation (tRNA, 5S rRNA). In higher eukaryotes, the Pol III transcriptome has expanded to include additional, recently evolved ncRNA species that effectively broaden the footprint of Pol III transcription to additional cellular activities. Newly evolved ncRNAs function as riboregulators of autophagy (vault), immune signaling cascades (nc886), and translation (Alu, BC200, snaR). Notably, upregulation of Pol III transcription is frequently observed in cancer, and multiple ncRNA species are linked to both cancer progression and poor survival outcomes among cancer patients. In this review, we outline the basic features and functions of the Pol III transcriptome, and the evidence for dysregulation and dysfunction for each ncRNA in cancer. When taken together, recurrent patterns emerge, ranging from shared functional motifs that include molecular scaffolding and protein sequestration, overlapping protein interactions, and immunostimulatory activities, to the biogenesis of analogous small RNA fragments and noncanonical miRNAs, augmenting the function of the Pol III transcriptome and further broadening its role in cancer. This article is categorized under: RNA in Disease and Development > RNA in Disease RNA Processing > Processing of Small RNAs RNA Interactions with Proteins and Other Molecules > Protein-RNA Interactions: Functional Implications.

Late Fusion Multiple Kernel Clustering With Proxy Graph Refinement.

Multiple kernel clustering (MKC) optimally utilizes a group of pre-specified base kernels to improve clustering performance. Among existing MKC algorithms, the recently proposed late fusion MKC methods demonstrate promising clustering performance in various applications and enjoy considerable computational acceleration. However, we observe that the kernel partition learning and late fusion processes are separated from each other in the existing mechanism, which may lead to suboptimal solutions and adversely affect the clustering performance. In this article, we propose a novel late fusion multiple kernel clustering with proxy graph refinement (LFMKC-PGR) framework to address these issues. First, we theoretically revisit the connection between late fusion kernel base partition and traditional spectral embedding. Based on this observation, we construct a proxy self-expressive graph from kernel base partitions. The proxy graph in return refines the individual kernel partitions and also captures partition relations in graph structure rather than simple linear transformation. We also provide theoretical connections and considerations between the proposed framework and the multiple kernel subspace clustering. An alternate algorithm with proved convergence is then developed to solve the resultant optimization problem. After that, extensive experiments are conducted on 12 multi-kernel benchmark datasets, and the results demonstrate the effectiveness of our proposed algorithm. The code of the proposed algorithm is publicly available at https://github.com/wangsiwei2010/graphlatefusion_MKC.

Multiple Kernel Clustering With Compressed Subspace Alignment.

Multiple kernel clustering (MKC) has recently achieved remarkable progress in fusing multisource information to boost the clustering performance. However, the O(n2) memory consumption and O(n3) computational complexity prohibit these methods from being applied into median- or large-scale applications, where n denotes the number of samples. To address these issues, we carefully redesign the formulation of subspace segmentation-based MKC, which reduces the memory and computational complexity to O(n) and O(n2) , respectively. The proposed algorithm adopts a novel sampling strategy to enhance the performance and accelerate the speed of MKC. Specifically, we first mathematically model the sampling process and then learn it simultaneously during the procedure of information fusion. By this way, the generated anchor point set can better serve data reconstruction across different views, leading to improved discriminative capability of the reconstruction matrix and boosted clustering performance. Although the integrated sampling process makes the proposed algorithm less efficient than the linear complexity algorithms, the elaborate formulation makes our algorithm straightforward for parallelization. Through the acceleration of GPU and multicore techniques, our algorithm achieves superior performance against the compared state-of-the-art methods on six datasets with comparable time cost to the linear complexity algorithms.

E 3Outlier: a Self-Supervised Framework for Unsupervised Deep Outlier Detection.

Existing unsupervised outlier detection (OD) solutions face a grave challenge with surging visual data like images. Although deep neural networks (DNNs) prove successful for visual data, deep OD remains difficult due to OD's unsupervised nature. This paper proposes a novel framework named E 3Outlier that can perform effective and end-to-end deep outlier removal. Its core idea is to introduce self-supervision into deep OD. Specifically, our major solution is to adopt a discriminative learning paradigm that creates multiple pseudo classes from given unlabeled data by various data operations, which enables us to apply prevalent discriminative DNNs (e.g., ResNet) to the unsupervised OD problem. Then, with theoretical and empirical demonstration, we argue that inlier priority, a property that encourages DNN to prioritize inliers during self-supervised learning, makes it possible to perform end-to-end OD. Meanwhile, unlike frequently-used outlierness measures (e.g., density, proximity) in previous OD methods, we explore network uncertainty and validate it as a highly effective outlierness measure, while two practical score refinement strategies are also designed to improve OD performance. Finally, in addition to the discriminative learning paradigm above, we also explore the solutions that exploit other learning paradigms (i.e., generative learning and contrastive learning) to introduce self-supervision for E 3Outlier. Such extendibility not only brings further performance gain on relatively difficult datasets, but also enables E 3Outlier to be applied to other OD applications like video abnormal event detection. Extensive experiments demonstrate that E 3Outlier can considerably outperform state-of-the-art counterparts by 10%-30% AUROC. Demo codes are available at https://github.com/demonzyj56/E3Outlier.

Deep Fusion Clustering Network With Reliable Structure Preservation.

Deep clustering, which can elegantly exploit data representation to seek a partition of the samples, has attracted intensive attention. Recently, combining auto-encoder (AE) with graph neural networks (GNNs) has accomplished excellent performance by introducing structural information implied among data in clustering tasks. However, we observe that there are some limitations of most existing works: 1) in practical graph datasets, there exist some noisy or inaccurate connections among nodes, which would confuse network learning and cause biased representations, thus leading to unsatisfied clustering performance; 2) lacking dynamic information fusion module to carefully combine and refine the node attributes and the graph structural information to learn more consistent representations; and 3) failing to exploit the two separated views' information for generating a more robust target distribution. To solve these problems, we propose a novel method termed deep fusion clustering network with reliable structure preservation (DFCN-RSP). Specifically, the random walk mechanism is introduced to boost the reliability of the original graph structure by measuring localized structure similarities among nodes. It can simultaneously filter out noisy connections and supplement reliable connections in the original graph. Moreover, we provide a transformer-based graph auto-encoder (TGAE) that can use a self-attention mechanism with the localized structure similarity information to fine-tune the fused topology structure among nodes layer by layer. Furthermore, we provide a dynamic cross-modality fusion strategy to combine the representations learned from both TGAE and AE. Also, we design a triplet self-supervision strategy and a target distribution generation measure to explore the cross-modality information. The experimental results on five public benchmark datasets reflect that DFCN-RSP is more competitive than the state-of-the-art deep clustering algorithms. The corresponding code is available at https://github.com/gongleii/DFCN-RSP.

Semantic instance segmentation with discriminative deep supervision for medical images.

Semantic instance segmentation is crucial for many medical image analysis applications, including computational pathology and automated radiation therapy. Existing methods for this task can be roughly classified into two categories: (1) proposal-based methods and (2) proposal-free methods. However, in medical images, the irregular shape-variations and crowding instances (e.g., nuclei and cells) make it hard for the proposal-based methods to achieve robust instance localization. On the other hand, ambiguous boundaries caused by the low-contrast nature of medical images (e.g., CT images) challenge the accuracy of the proposal-free methods. To tackle these issues, we propose a proposal-free segmentation network with discriminative deep supervision (DDS), which at the same time allows us to gain the power of the proposal-based method. The DDS module is interleaved with a carefully designed proposal-free segmentation backbone in our network. Consequently, the features learned by the backbone network become more sensitive to instance localization. Also, with the proposed DDS module, robust pixel-wise instance-level cues (especially structural information) are introduced for semantic segmentation. Extensive experiments on three datasets, i.e., a nuclei dataset, a pelvic CT image dataset, and a synthetic dataset, demonstrate the superior performance of the proposed algorithm compared to the previous works.

Interpolation-Based Contrastive Learning for Few-Label Semi-Supervised Learning.

Semi-supervised learning (SSL) has long been proved to be an effective technique to construct powerful models with limited labels. In the existing literature, consistency regularization-based methods, which force the perturbed samples to have similar predictions with the original ones have attracted much attention for their promising accuracy. However, we observe that the performance of such methods decreases drastically when the labels get extremely limited, e.g., 2 or 3 labels for each category. Our empirical study finds that the main problem lies with the drift of semantic information in the procedure of data augmentation. The problem can be alleviated when enough supervision is provided. However, when little guidance is available, the incorrect regularization would mislead the network and undermine the performance of the algorithm. To tackle the problem, we: 1) propose an interpolation-based method to construct more reliable positive sample pairs and 2) design a novel contrastive loss to guide the embedding of the learned network to change linearly between samples so as to improve the discriminative capability of the network by enlarging the margin decision boundaries. Since no destructive regularization is introduced, the performance of our proposed algorithm is largely improved. Specifically, the proposed algorithm outperforms the second best algorithm (Comatch) with 5.3% by achieving 88.73% classification accuracy when only two labels are available for each class on the CIFAR-10 dataset. Moreover, we further prove the generality of the proposed method by improving the performance of the existing state-of-the-art algorithms considerably with our proposed strategy. The corresponding code is available at https://github.com/xihongyang1999/ICL_SSL.