Node Labels Research Articles

Colorectal cancer lymph node metastasis prediction with weakly supervised transformer-based multi-instance learning

Lymph node metastasis examined by the resected lymph nodes is considered one of the most important prognostic factors for colorectal cancer (CRC). However, it requires careful and comprehensive inspection by expert pathologists. To relieve the pathologists’ burden and speed up the diagnostic process, in this paper, we develop a deep learning system with the binary positive/negative labels of the lymph nodes to solve the CRC lymph node classification task. The multi-instance learning (MIL) framework is adopted in our method to handle the whole slide images (WSIs) of gigapixels in size at once and get rid of the labor-intensive and time-consuming detailed annotations. First, a transformer-based MIL model, DT-DSMIL, is proposed in this paper based on the deformable transformer backbone and the dual-stream MIL (DSMIL) framework. The local-level image features are extracted and aggregated with the deformable transformer, and the global-level image features are obtained with the DSMIL aggregator. The final classification decision is made based on both the local and the global-level features. After the effectiveness of our proposed DT-DSMIL model is demonstrated by comparing its performance with its predecessors, a diagnostic system is developed to detect, crop, and finally identify the single lymph nodes within the slides based on the DT-DSMIL and the Faster R-CNN model. The developed diagnostic model is trained and tested on a clinically collected CRC lymph node metastasis dataset composed of 843 slides (864 metastasis lymph nodes and 1415 non-metastatic lymph nodes), achieving the accuracy of 95.3% and the area under the receiver operating characteristic curve (AUC) of 0.9762 (95% confidence interval [CI]: 0.9607–0.9891) for the single lymph node classification. As for the lymph nodes with micro-metastasis and macro-metastasis, our diagnostic system achieves the AUC of 0.9816 (95% CI: 0.9659–0.9935) and 0.9902 (95% CI: 0.9787–0.9983), respectively. Moreover, the system shows reliable diagnostic region localizing performance: the model can always identify the most likely metastases, no matter the model’s predictions or manual labels, showing great potential in avoiding false negatives and discovering incorrectly labeled slides in actual clinical use.Graphical

Density of states for fast embedding node-attributed graphs

Given a node-attributed graph, how can we efficiently represent it with few numerical features that expressively reflect its topology and attribute information? We propose A-DOGE, for attributed DOS-based graph embedding, based on density of states (DOS, a.k.a. spectral density) to tackle this problem. A-DOGE is designed to fulfill a long desiderata of desirable characteristics. Most notably, it capitalizes on efficient approximation algorithms for DOS, that we extend to blend in node labels and attributes for the first time, making it fast and scalable for large attributed graphs and graph databases. Being based on the entire eigenspectrum of a graph, A-DOGE can capture structural and attribute properties at multiple (“glocal”) scales. Moreover, it is unsupervised (i.e., agnostic to any specific objective) and lends itself to various interpretations, which makes it suitable for exploratory graph mining tasks. Finally, it processes each graph independent of others, making it amenable for streaming settings as well as parallelization. Through extensive experiments, we show the efficacy and efficiency of A-DOGE on exploratory graph analysis and graph classification tasks, where it significantly outperforms unsupervised baselines and achieves competitive performance with modern supervised GNNs, while achieving the best trade-off between accuracy and runtime.

Revisiting Homophily Ratio: A Relation-Aware Graph Neural Network for Homophily and Heterophily

The graph neural network (GNN) is a type of powerful deep learning model used to process graph data consisting of nodes and edges. Many studies of GNNs have modeled the relationships between the edges and labels of nodes only by homophily/heterophily, where most/few nodes with the same label tend to have an edge between each other. However, this modeling method cannot describe the multiconnection mode on graphs where homophily can coexist with heterophily. In this work, we propose a transition matrix to describe the relationships between edges and labels at the class level. Through this transition matrix, we constructed a more interpretable GNN in a neighbor-predicting manner, measured the information that the edges can provide for the node classification task, and proposed a method to test whether the labels match the edges. The results show the improvement of the proposed method against state-of-the-art (SOTA) GNNs. We also obtain the following two results: (1) the poor performance of GNNs is highly relevant to the information of edges instead of heterophily, which is always considered the main factor resulting in the decline in performance; and (2) most benchmark heterophilic datasets exhibit the label-edge mismatch problem, leading them to become intractable

HINChip: Heterogeneous Information Network Representation with Community Hierarchy Preserving

Heterogeneous information network (HIN) representation aims to learn low-dimensional representations for multiple types of nodes and edges while preserving rich structural and semantic information. Existing HIN embedding methods mainly utilize meta-path, meta-graph, and network schema to guide the learned representations. However, these methods may ignore the community hierarchy information in the information network, which is commonly available in many applications. In this paper, we propose HINChip, a new approach for learning representations of Heterogeneous Information Networks with Community Hierarchy Preserving, including intra-community structures and inter-community relationships. Specifically, multiple homogeneous sub-networks are first constructed based on multiple semantics. Second, hierarchical networks are built from fine to coarse on each homogeneous sub-network, preserving intra-community structures and inter-community relationships on the corresponding semantics. The nodes learn the corresponding vectors from each level of the hierarchical networks by the existing representation learning method. Third, multi-level-integrated semantic attention is proposed to combine the node vectors of each level and weigh the importance of different homogeneous sub-networks under given semantics. The proposed HINChip considers only the topology of HIN and outperforms methods trained with node attributes and labels in the experiments conducted on four real-world datasets.

Feature selection: Key to enhance node classification with graph neural networks

AbstractGraphs help to define the relationships between entities in the data. These relationships, represented by edges, often provide additional context information which can be utilised to discover patterns in the data. Graph Neural Networks (GNNs) employ the inductive bias of the graph structure to learn and predict on various tasks. The primary operation of graph neural networks is the feature aggregation step performed over neighbours of the node based on the structure of the graph. In addition to its own features, for each hop, the node gets additional combined features from its neighbours. These aggregated features help define the similarity or dissimilarity of the nodes with respect to the labels and are useful for tasks like node classification. However, in real‐world data, features of neighbours at different hops may not correlate with the node's features. Thus, any indiscriminate feature aggregation by GNN might cause the addition of noisy features leading to degradation in model's performance. In this work, we show that selective aggregation of node features from various hops leads to better performance than default aggregation on the node classification task. Furthermore, we propose a Dual‐Net GNN architecture with a classifier model and a selector model. The classifier model trains over a subset of input node features to predict node labels while the selector model learns to provide optimal input subset to the classifier for the best performance. These two models are trained jointly to learn the best subset of features that give higher accuracy in node label predictions. With extensive experiments, we show that our proposed model outperforms both feature selection methods and state‐of‐the‐art GNN models with remarkable improvements up to 27.8%.

Offline EEG hyper-scanning using anonymous walk embeddings in tacit coordination games.

In this paper we present a method to examine the synchrony between brains without the need to carry out simultaneous recordings of EEG signals from two people which is the essence of hyper-scanning studies. We used anonymous random walks to spatially encode the entire graph structure without relying on data at the level of individual nodes. Anonymous random walks enabled us to encapsulate the structure of a graph regardless of the specific node labels. That is, random walks that visited different nodes in the same sequence resulted in the same anonymous walk encoding. We have analyzed the EEG data offline and matched each possible pair of players from the entire pool of players that performed a series of tacit coordination games. Specifically, we compared between two network patterns associated with each possible pair of players. By using classification performed on the spatial distance between each pair of individual brain patterns, analyzed by the random walk algorithm, we tried to predict whether each possible pair of players has managed to converge on the same solution in each tacit coordination game. Specifically, the distance between a pair of vector embeddings, each associated with one of the players, was used as input for a classification model for the purpose of predicting whether the two corresponding players have managed to achieve successful coordination. Our model reached a classification accuracy of ~85%.

Tree trace reconstruction using subtraces

AbstractTree trace reconstruction aims to learn the binary node labels of a tree, given independent samples of the tree passed through an appropriately defined deletion channel. In recent work, Davies, Rácz, and Rashtchian [10] used combinatorial methods to show that $\exp({\mathrm{O}} (k \log_{k} n))$ samples suffice to reconstruct a complete k-ary tree with n nodes with high probability. We provide an alternative proof of this result, which allows us to generalize it to a broader class of tree topologies and deletion models. In our proofs we introduce the notion of a subtrace, which enables us to connect with and generalize recent mean-based complex analytic algorithms for string trace reconstruction.

Universal and Succinct Source Coding of Deep Neural Networks

Deep neural networks have shown incredible performance for inference tasks in a variety of domains, but require significant storage space, which limits scaling and use for on-device intelligence. This paper is concerned with finding universal lossless compressed representations of deep feedforward networks with synaptic weights drawn from discrete sets, and directly performing inference without full decompression. The basic insight that allows less rate than naïve approaches is recognizing that the bipartite graph layers of feedforward networks have a kind of permutation invariance to the labeling of nodes, in terms of inferential operation. We provide efficient algorithms to dissipate this irrelevant uncertainty and then use arithmetic coding to nearly achieve the entropy bound in a universal manner. We also provide experimental results of our approach on several standard datasets.

Semi-Automatic Defect Distinction of PAUT for High Strength Steel Corner Structure

In this study, A GCN distinction network is proposed to identify the defect status of steel corner structures. In this defect determination method, the detected S-scan data is converted into A-scan by Doppler software to form a two-dimensional detection matrix, and the two dimensions are deflection angle stepping and sampling point on the acoustic distance respectively. GCN needs data node relations for GCN’s adjacent matrix, which concludes the shape of the sample and the reflection information and the structure of acoustic beam direction. The conventional inverse and threshold method are introduced in order to enhance classification performance. According to the node and the adjacency matrix relationship between nodes of affiliation and node sampling value automatically set the node label, The GCN model is constructed. Due to the characteristics of GCN, the beam length in the adjacent matrix cannot be changed again in the application test. Except for the detection model of Beamtool software, all other data processes and distinction are carried out automatically. In this study, the PAUT of low-alloy steel (30CrMnSi) was carried out by the PHASCAN II detector, to verify the effectiveness of the research method through natural wave and defect echo experiments which will be confused to distinguish.

Temporal burstiness and collaborative camouflage aware fraud detection

With the prosperity and development of the digital economy, many fraudsters have emerged on e-commerce platforms to fabricate fraudulent reviews to mislead consumers’ shopping decisions for profit. Moreover, in order to evade fraud detection, fraudsters continue to evolve and present the phenomenon of adversarial camouflage and collaborative attack. In this paper, we propose a novel temporal burstiness and collaborative camouflage aware method (TBCCA) for fraudster detection. Specifically, we capture the hidden temporal burstiness features behind camouflage strategy based on the time series prediction model, and identify highly suspicious target products by assigning suspicious scores as node priors. Meanwhile, a propagation graph integrating review collusion is constructed, and an iterative fraud confidence propagation algorithm is designed for inferring the label of nodes in the graph based on Loop Belief Propagation (LBP). Comprehensive experiments are conducted to compare TBCCA with state-of-the-art fraudster detection approaches, and experimental results show that TBCCA can effectively identify fraudsters in real review networks with achieving 6%–10% performance improvement than other baselines.

A graph neural network framework for mapping histological topology in oral mucosal tissue

BackgroundHistological feature representation is advantageous for computer aided diagnosis (CAD) and disease classification when using predictive techniques based on machine learning. Explicit feature representations in computer tissue models can assist explainability of machine learning predictions. Different approaches to feature representation within digital tissue images have been proposed. Cell-graphs have been demonstrated to provide precise and general constructs that can model both low- and high-level features. The basement membrane is high-level tissue architecture, and interactions across the basement membrane are involved in multiple disease processes. Thus, the basement membrane is an important histological feature to study from a cell-graph and machine learning perspective.ResultsWe present a two stage machine learning pipeline for generating a cell-graph from a digital H &E stained tissue image. Using a combination of convolutional neural networks for visual analysis and graph neural networks exploiting node and edge labels for topological analysis, the pipeline is shown to predict both low- and high-level histological features in oral mucosal tissue with good accuracy.ConclusionsConvolutional and graph neural networks are complementary technologies for learning, representing and predicting local and global histological features employing node and edge labels. Their combination is potentially widely applicable in histopathology image analysis and can enhance explainability in CAD tools for disease prediction.

Similarity learning with deep CRF for person re-identification

The core of person re-identification (Re-ID) lies in robustly estimating similarities for each probe-gallery image pair. A common practice in existing works is to calculate the similarity of each image pair independently, ignoring relations between different probe-gallery pairs. In this paper, we present a deep learning conditional random field (Deep-CRF) graph to model group-wise similarities within a batch of images, and regard the Re-ID task as a CRF node labeling problem. Unlike the existing deep CRF based approach where the CRF inference is only involved in the training stage, our method intends to fully exploit the potential of CRF model, exhibiting inference consistency in both training and testing. Specifically, we design unary potentials for computing each probe-gallery similarity separately. To efficiently encode relationships between different probe-gallery pairs, pairwise potentials are built on an arbitrary node pair whose learning is achieved by a joint matching strategy using bidirectional LSTM. We pose the CRF inference as a RNN learning process, where unary and pairwise potentials are jointly optimized in an end-to-end manner. Extensive experiments on three large-scale person Re-ID datasets demonstrate the effectiveness of the proposed method. Our Deep-CRF achieves the best results compared with the previous graph-based deep learning approaches and substantially exceeds the existing deep CRF framework by 8% in Rank1 accuracy on CUHK03 dataset. It also behaves competitive among the current state-of-the-art methods.

Clinical Decision Support through Interpretable Machine Learning in Head and Neck Cancer

<h3>Purpose/Objective(s)</h3> Identifying extracapsular extension (ECE) in Head and neck squamous cell carcinoma (HNSCC) is critical to accurate staging and treatment selection. Still, it currently cannot be known accurately until pathologic analysis post-surgery. Guided by routine computed tomography (CT) scans, artificial intelligence can be a high-potential approach to assess tumors and help with ECE identification before surgery. In this research, we achieve automatic ECE identification on CT by proposing a novel gradient-based class activation mapping (Grad-CAM) model. The Grad-CAM model can guide the machine learning process to focus on highly suspicious ECE regions without lymph node label information. This research goes a step beyond prediction accuracy and into the realm of clinical decision support by demonstrating interpretable implementation via a "heatmap" of anatomy relevant to ECE visible to the end-user. <h3>Materials/Methods</h3> The studied dataset includes 130 patients' CT scans, including ECE positive and negative patients. We propose an interpretable deep neural network framework to perform ECE identification without requiring annotated lymph node regions. Because Grad-CAM can provide visual explanations for deep neural networks by indicating the critical regions in the image for prediction, this technique is applied to guide the learning algorithm and generate "heatmaps," which can highlight the highly related regions ECE. Informative volumes of interest (VOIs) are extracted based on the heatmaps, and a classifier is trained upon the VOIs to perform ECE classification. The generated "heatmaps" are then compared with the pathology to demonstrate potential clinical decision support. <h3>Results</h3> The experimental results have demonstrated that our model can identify ECE positive and negative patients. In evaluation, the proposed method has achieved test accuracy and AUC of 90.2% and 91.1%, respectively. ECE probability heatmaps are compared with pathological lymph node extension in Table 1. <h3>Conclusion</h3> This study demonstrates that our proposed machine learning model guided by the Grad-CAM approach can be used for ECE identification. The proposed deep neural networks can help capture sufficient features and achieve high classification accuracy. The "heatmaps" can be used as a tool for clinicians to identify ECE and guide clinical decisions. The outcome of this research is expected to promote the implementation of interpretable artificial intelligence for HN cancer ECE identification in clinical practice.

SFTM: Fast matching of web pages using Similarity-based Flexible Tree Matching

Tree matching techniques have been investigated in many fields, including web data mining and extraction, as a key component to analyze the content of web pages. However, when applied to existing web pages, traditional tree matching approaches, covered by algorithms like Tree-Edit Distance (TED) or XyDiff, either fail to scale beyond a few hundred nodes or exhibit a relatively low accuracy.In this article, we therefore propose a novel algorithm, named Similarity-based Flexible Tree Matching (SFTM), which enables high accuracy tree matching on real-life web pages, with practical computation times. We approach tree matching as an optimization problem and leverage node labels and local topology similarity in order to avoid any combinatorial explosion. Our practical evaluation demonstrates that SFTM significantly improves the state of the art in terms of accuracy, while allowing computation times significantly lower than the most accurate solutions. By gaining on these two dimensions, SFTM therefore offers an affordable solution to match complex trees in practice.

Multiplex network infomax: Multiplex network embedding via information fusion

For networking of big data applications, an essential issue is how to represent networks in vector space for further mining and analysis tasks, e.g., node classification, clustering, link prediction, and visualization. Most existing studies on this subject mainly concentrate on monoplex networks considering a single type of relation among nodes. However, numerous real-world networks are naturally composed of multiple layers with different relation types; such a network is called a multiplex network. The majority of existing multiplex network embedding methods either overlook node attributes, resort to node labels for training, or underutilize underlying information shared across multiple layers. In this paper, we propose Multiplex Network Infomax (MNI), an unsupervised embedding framework to represent information of multiple layers into a unified embedding space. To be more specific, we aim to maximize the mutual information between the unified embedding and node embeddings of each layer. On the basis of this framework, we present an unsupervised network embedding method for attributed multiplex networks. Experimental results show that our method achieves competitive performance on not only node-related tasks, such as node classification, clustering, and similarity search, but also a typical edge-related task, i.e., link prediction, at times even outperforming relevant supervised methods, despite that MNI is fully unsupervised.

Building Function Recognition Using the Semi-Supervised Classification

The functional classification of buildings is important for creating and managing urban zones and assisting government departments. Building function recognition is incredibly valuable for wide applications ranging from the determination of energy demand. By aiming at the topic of urban function classification, a semi-supervised graph structure network combined unified message passing model was introduced. The data of this model include spatial location distribution of buildings, building characteristics and the information mined from points of interesting (POIs). In order to extract the context information, each building was regarded as a graph node. Building characteristics and corresponding POIs information were embedded to mine the building function by the graph convolutional neural network. When training the model, several node labels in the graph were masked, and then these labels were predicted by the trained model so that this work could take full advantage of the node label and the feature information of all nodes in both the training and prediction stages. Quasi-experiments proved that the proposed method for building function classification using multi-source data enables the model to capture more meaningful information with limited labels, and it achieves better function classification results.

Robust rotating machinery diagnosis using a dynamic-weighted graph updating strategy

Recently, graph data analysis receives attentions in mechanical fault diagnosis. Edge connection of the input graph indicates that neighbor nodes share the same fault type, but differences in importance between neighbor nodes is rarely demonstrated. Also, noise is unavoidable during signal acquisition, affecting the quality and reliability of constructed input graphs. In this paper, a robust rotating machinery diagnosis using a dynamic-weighted graph updating strategy is presented. Firstly, a non-weighted graph is constructed using the energy spectrum and the distance metric. Second, a linear weighting strategy is developed to assign weights to the edge connections, adjusting the influence of neighbor nodes quantitatively. Then, a dynamic edge-weight updating process is developed to decrease the negative influence of noisy samples using high-level output features during model training. Finally, node labels responding to the fault types of samples are identified. Case studies verified the effectiveness of the proposed method with noisy samples.

Heri-Graphs: A Dataset Creation Framework for Multi-Modal Machine Learning on Graphs of Heritage Values and Attributes with Social Media

Values (why to conserve) and Attributes (what to conserve) are essential concepts of cultural heritage. Recent studies have been using social media to map values and attributes conveyed by the public to cultural heritage. However, it is rare to connect heterogeneous modalities of images, texts, geo-locations, timestamps, and social network structures to mine the semantic and structural characteristics therein. This study presents a methodological framework for constructing such multi-modal datasets using posts and images on Flickr for graph-based machine learning (ML) tasks concerning heritage values and attributes. After data pre-processing using pre-trained ML models, the multi-modal information of visual contents and textual semantics are modelled as node features and labels, while their social relationships and spatio-temporal contexts are modelled as links in Multi-Graphs. The framework is tested in three cities containing UNESCO World Heritage properties—Amsterdam, Suzhou, and Venice— which yielded datasets with high consistency for semi-supervised learning tasks. The entire process is formally described with mathematical notations, ready to be applied in provisional tasks both as ML problems with technical relevance and as urban/heritage study questions with societal interests. This study could also benefit the understanding and mapping of heritage values and attributes for future research in global cases, aiming at inclusive heritage management practices. Moreover, the proposed framework could be summarized as creating attributed graphs from unstructured social media data sources, ready to be applied in a wide range of use cases.

NIR-II and visible fluorescence hybrid imaging-guided surgery via aggregation-induced emission fluorophores cocktails.

Fluorescence imaging-guided surgery is one of important techniques to realize precision surgery. Although second near-infrared window (NIR-II) fluorescence imaging has the advantages of high resolution and large penetration depth in surgical navigation, its major drawback is that NIR-II images cannot be detected by our naked eyes, which demands a high hand-eye coordination for surgeons and increases the surgical difficulty. On the contrary, visible fluorescence can be observed by our naked eyes but has poor penetration. Here, we firstly propose a kind of NIR-II and visible fluorescence hybrid navigation surgery assisted via a cocktail of aggregation-induced emission nanoparticles (AIE NPs). NIR-II imaging helps to locate deep targeted tissues and judge the residual, and visible fluorescence offers an easily surgical navigation. We apply this hybrid navigation mode in different animals and systems, and verify that it can accelerate surgical process and compatible with a visible fluorescence endoscopy. To deepen the understanding of lymph node (LN) labelling, the distribution of NPs in LNs after local administration is initially analyzed by NIR-II fluorescence wide-filed microscopy, and two fates of the NPs are summarized. An alternative strategy which combines indocyanine green and berberine is also reported as a compromise for rapidly clinical translation.

A Gaze into the Internal Logic of Graph Neural Networks, with Logic

Graph Neural Networks share with Logic Programming several key relational inference mechanisms. The datasets on which they are trained and evaluated can be seen as database facts containing ground terms. This makes possible modeling their inference mechanisms with equivalent logic programs, to better understand not just how they propagate information between the entities involved in the machine learning process but also to infer limits on what can be learned from a given dataset and how well that might generalize to unseen test data. This leads us to the key idea of this paper: modeling with the help of a logic program the information flows involved in learning to infer from the link structure of a graph and the information content of its nodes properties of new nodes, given their known connections to nodes with possibly similar properties. The problem is known as graph node property prediction and our approach will consist in emulating with help of a Prolog program the key information propagation steps of a Graph Neural Network's training and inference stages. We test our a approach on the ogbn-arxiv node property inference benchmark. To infer class labels for nodes representing papers in a citation network, we distill the dependency trees of the text associated to each node into directed acyclic graphs that we encode as ground Prolog terms. Together with the set of their references to other papers, they become facts in a database on which we reason with help of a Prolog program that mimics the information propagation in graph neural networks predicting node properties. In the process, we invent ground term similarity relations that help infer labels in the test set by propagating node properties from similar nodes in the training set and we evaluate their effectiveness in comparison with that of the graph's link structure. Finally, we implement explanation generators that unveil performance upper bounds inherent to the dataset. As a practical outcome, we obtain a logic program, that, when seen as machine learning algorithm, performs close to the state of the art on the node property prediction benchmark.