Can a vision model truly see an object, or does it only fit surface-level visual cues? Following Wittgenstein's view that the limits of language are the limits of the world, we view a model's recognition ability as bounded by the descriptive system it has learned. In current vision models, this system is often realized through learned feature representations that exploit local statistical cues. We therefore ask whether a model can still classify correctly when such local cues provide no stable basis for distinction. We formalize this question with syntactic distance, which measures class separability through the symmetry of the operations mapping one class to the other: positive distance exposes exploitable local features, whereas zero distance requires global semantics rather than local rules. We construct a visual self-referential task in maximum-variance binary noise: positive samples contain a closed square, while negative samples contain an otherwise identical square with one flipped boundary pixel. The two classes differ in global semantics but have zero syntactic distance, making local statistical shortcuts unreliable. Experiments on ResNets and Vision Transformers reveal a consistent phase-transition phenomenon, with accuracy collapsing to random guessing once the image scale crosses a critical point and does not recover within the tested range. Larger training sets and models only delay this collapse, while globally attentive ViTs reach it earlier. These results reveal a structural capability boundary of current architectures on global-concept tasks, suggesting that general intelligence may require creating new language, not reusing an existing one.

Trapped by the label scarcity in molecular property prediction and drug design, graph contrastive learning (GCL) came forward. Leading contrastive learning works show two kinds of view generators, that is, random or learnable data corruption and domain knowledge incorporation. While effective, the two ways also lead to molecular semantics altering and limited generalization capability, respectively. To this end, we relate the LinE graph with MOlecular graph coNtrastive learning and propose a novel method termed LEMON. Specifically, by contrasting the given graph with the corresponding line graph, the graph encoder can freely encode the molecular semantics without omission. Furthermore, we present a new patch with edge attribute fusion and two local contrastive losses enhance information transmission and tackle hard negative samples. Compared with state-of-the-art (SOTA) methods for view generation, superior performance on molecular property prediction suggests the effectiveness of our proposed framework.

ABSTRACT Technical analysis, which includes technical indicators and charts derived from specific rules, has proven effective and widely used for stock movement prediction. However, technical chart evaluation is often limited by subjectivity, arising from sparse chart types and substantial information loss due to rigid rules. While pattern recognition algorithms have been developed to address this issue, they still rely on manual chart labelling and primarily focus on closing prices, leaving much of the chart's broader information untapped. To overcome these limitations, we propose a novel framework called ChartNet, designed to extract general information from technical charts and reduce subjectivity in chart analysis. ChartNet employs a unified representation for charts across financial series with varying simplification levels and leverages a chart triplet loss function for unsupervised training, eliminating the need for labelled data. Compared with several state‐of‐the‐art baselines, our framework has reached the best prediction accuracy on CSI‐300, SZ‐50 components and Dow Jones Index in 2022: 65.91%, 63.70% and 64.96% respectively. In backtesting using actual stock data, our framework achieves the highest average return of 1.12 and 1.15. Furthermore, we highlight the interpretability of ChartNet through two case studies, some important charts and failure cases, illustrating its capability to uncover meaningful insights from charts. This research contributes to advancing the objective evaluation of technical charts and promoting a more comprehensive understanding of chart‐based stock prediction performance.

Existing self-supervised methods in natural language processing (NLP), especially hierarchical text classification (HTC), mainly focus on self-supervised contrastive learning, extremely relying on human-designed augmentation rules to generate contrastive samples, which can potentially corrupt or distort the original information. In this paper, we tend to investigate the feasibility of a contrastive learning scheme in which the semantic and syntactic information inherent in the input sample is adequately reserved in the contrastive samples and fused during the learning process. Specifically, we propose an information lossless contrastive learning strategy for HTC, namely Hierarchy-aware Information Lossless contrastive Learning (HILL), which consists of a text encoder representing the input document, and a structure encoder directly generating the positive sample. The structure encoder takes the document embedding as input, extracts the essential syntactic information inherent in the label hierarchy with the principle of structural entropy minimization, and injects the syntactic information into the text representation via hierarchical representation learning. Experiments on three common datasets are conducted to verify the superiority of HILL.

Forecasting financial data plays a crucial role in financial market. Relying solely on prices or price trends as prediction targets often leads to a vast of invalid transactions. As a result, researchers have increasingly turned their attention to turning points as the prediction target. Surprisingly, existing methods have largely overlooked the role of technical charts, despite turning points being closely related to the technical charts. Recently, several researchers have attempted to utilize chart information via converting price sequences into images for turning point forecasting, but robustness and convergence problems arise. To address these challenges and enhance the turning point predictions, this article introduces a new method known as MPCNet. Specifically, we first transform the price series into a graph structure using chart similarity to robustly extract valuable information from technical charts. Additionally, we introduce the multipersistence topology tool to accurately predict stock turning points and provide convergence guarantee. Experimental results demonstrate the significant superiority of our proposed model over existing methods. Furthermore, based on additional performance evaluations using real stock data, MPCNet consistently achieves the highest average return during the transaction backtesting period. Meanwhile, we provide empirical validation of robustness and theoretical analysis to confirm its convergence, establishing it as a superior tool for financial forecasting.

Due to the success observed in deep neural networks with contrastive learning, there has been a notable surge in research interest in graph contrastive learning, primarily attributed to its superior performance in graphs with limited labeled data. Within contrastive learning, the selection of a “view” dictates the information captured by the representation, thereby influencing the model’s performance. However, assessing the quality of information in these views poses challenges, and determining what constitutes a good view remains unclear. This paper addresses this issue by establishing the definition of a good view through the application of graph information bottleneck and structural entropy theories. Based on theoretical insights, we introduce CtrlGCL, a novel method for achieving a beneficial view in graph contrastive learning through coding tree representation learning. Extensive experiments were conducted to ascertain the effectiveness of the proposed view in unsupervised and semi-supervised learning. In particular, our approach, via CtrlGCL-H, yields an average accuracy enhancement of 1.06% under unsupervised learning when compared to GCL. This improvement underscores the efficacy of our proposed method.

Thanks to the great success of graph neural networks (GNNs) in structural information learning, extensive variants by virtue of sampling or pooling have been developed to further improve the performance, scalability, and applicability. However, there is still room for improvement in learning efficiency because current GNNs are trained via batch gradient descent with many graphs in each iteration. The good potential of random features in speeding up the training phase motivates us to consider the expressive power of GNNs with random weights. Based on the framework of Graph Isomorphism Network, we propose a novel model called Hashing Graph Isomorphism Network (HashGIN) with only one epoch of training by revising the convolutional layer with random hash functions and adjusting the learning objective with regularized least squares loss. In light of the property of $k$ random hash functions, we theoretically show that the injective phase in the Weisfeiler-Lehman test can be approximated by a hash family. An approximation upper bound is further provided with rigorous mathematical proof for the convergence of our model. Our experiments on several benchmark datasets show that HashGIN is effective and efficient for graph classification tasks. Compared to the state-of-the-art methods, HashGIN achieves better or comparable accuracies with less training time and memory cost.

Graph contrastive learning has achieved great success in pre-training graph neural networks without ground-truth labels. Leading graph contrastive learning follows the classical scheme of contrastive learning, forcing model to identify the essential information from augmented views. However, general augmented views are produced via random corruption or learning, which inevitably leads to semantics alteration. Although domain knowledge guided augmentations alleviate this issue, the generated views are domain specific and undermine the generalization. In this work, motivated by the firm representation ability of sparse model from pruning, we reformulate the problem of graph contrastive learning via contrasting different model versions rather than augmented views. We first theoretically reveal the superiority of model pruning in contrast to data augmentations. In practice, we take original graph as input and dynamically generate a perturbed graph encoder to contrast with the original encoder by pruning its transformation weights. Furthermore, considering the integrity of node embedding in our method, we are capable of developing a local contrastive loss to tackle the hard negative samples that disturb the model training. We extensively validate our method on various benchmarks regarding graph classification via unsupervised and transfer learning. Compared to the state-of-the-art (SOTA) works, better performance can always be obtained by the proposed method.

Given the pivotal role of molecular property prediction in drug development and material science, graph self-supervised learning has been implemented in molecular representation learning to compensate for the shortage of labeled molecules. However, current proposed methods often focus on designing data augmentation schemes and leveraging domain knowledge to improve performance, which inevitably leads to molecular semantics loss and limited generalization capability. To the end, we propose IPM, an Information lossless Pretraining strategy for Molecular property prediction that leverages the information of both the original graph and line graph of molecules. Specifically, by contrasting the given graph with the corresponding line graph, the graph encoder can fully learn the generic molecular semantic representation without profound domain knowledge. We also design a new message-passing scheme that retains information consistency during message passing between two kinds of graphs. Additionally, we present two graph contrastive losses for performance fixing and over-smoothing prevention during the learning process. Experimental results on multiple regression tasks for molecular property prediction demonstrate the effectiveness of IPM against state-of-the-art (SOTA) methods.

Trapped by the label scarcity in molecular property prediction and drug design, graph contrastive learning came forward. A general contrastive model consists of a view generator, view encoder, and contrastive loss, in which the view mainly controls the encoded information underlying input graphs. Leading contrastive learning works show two kinds of view generators, that is, random or learnable data corruption and domain knowledge incorporation. While effective, the two ways also lead to molecular semantics altering and limited generalization capability, respectively. Thus, a decent view that can fully retain molecular semantics and is free from profound domain knowledge is supposed to come forward. To this end, we relate molecular graph contrastive learning with the line graph and propose a novel method termed LGCL. Specifically, by contrasting the given graph with the corresponding line graph, the graph encoder can freely encode the molecular semantics without omission. While considering the information inconsistency and over-smoothing derived from the learning process because of the mismatched pace of message passing in two kinds of graphs, we present a new patch with edge attribute fusion and two local contrastive losses for performance fixing. Compared with state-of-the-art (SOTA) methods for view generation, superior performance on molecular property prediction suggests the effectiveness of line graphs severing as the contrasting views.

In contrastive learning, the choice of ``view'' controls the information that the representation captures and influences the performance of the model. However, leading graph contrastive learning methods generally produce views via random corruption or learning, which could lead to the loss of essential information and alteration of semantic information. An anchor view that maintains the essential information of input graphs for contrastive learning has been hardly investigated. In this paper, based on the theory of graph information bottleneck, we deduce the definition of this anchor view; put differently, \textit{the anchor view with essential information of input graph is supposed to have the minimal structural uncertainty}. Furthermore, guided by structural entropy, we implement the anchor view, termed \textbf{SEGA}, for graph contrastive learning. We extensively validate the proposed anchor view on various benchmarks regarding graph classification under unsupervised, semi-supervised, and transfer learning and achieve significant performance boosts compared to the state-of-the-art methods.

Great research efforts have been devoted to exploiting deep neural networks in stock prediction. However, long-term dependencies and chaotic properties are still two major issues that lower the performance of state-of-the-art deep learning models in forecasting future price trends. In this study, we propose a novel framework to address both issues. Specifically, in terms of transforming time series into complex networks, we convert market price series into graphs. Then, structural information, referring to temporal point associations and node weights, is extracted from the mapped graphs to resolve the problems regarding long-term dependencies and chaotic properties. We take graph embeddings to represent the associations among temporal points as the prediction model inputs. Node weights are used as a priori knowledge to enhance the learning of temporal attention. The effectiveness of our proposed framework is validated using real-world stock data, and our approach obtains the best performance among several state-of-the-art benchmarks. Moreover, in the conducted trading simulations, our framework further obtains the highest cumulative profits. Our results supplement the existing applications of complex network methods in the financial realm and provide insightful implications for investment applications regarding decision support in financial markets.

Advanced deep learning methods have been widely adopted in stock movement prediction with technical analysis (TA), while researchers prefer technical indicators to technical charts due to the divergence in quantification difficulty. In traditional TA, researchers usually utilize chart similarity to solve the quantifying problem, while chart similarity is often limited to specific charts as the templates for comparison, resulting in massive inadequate use of information. Accordingly, we propose a novel similarity framework to overcome the limitation of chart similarity to specific charts. Specifically, after extracting the key point sequence (i.e., the draft chart) from the stock price series, we transform it into a graph and ultimately employ an arbitrary graph kernel such as the Weisfeiler-Lehman graph kernel and graph convolutional network (GCN) to sufficiently mine the information in the chart for stock movement prediction. Our similarity framework is more robust than the chart similarity measures commonly used in traditional TA. Additionally, we further evaluate the effectiveness of our framework on real-world stock data and show that our framework achieves the best performance compared to several state-of-the-art baselines in stock movement prediction and obtains the highest average net values in a trading simulation. Our results complement the existing application of the chart similarity method in deep learning and provide support for the investing application of financial market decisions. (C)& nbsp;2022 Elsevier B.V. All rights reserved.

Following the success of convolution on non-Euclidean space, the corresponding pooling approaches have also been validated on various tasks regarding graphs. However, because of the fixed compression quota and stepwise pooling design, these hierarchical pooling methods still suffer from local structure damage and suboptimal problem. In this work, inspired by structural entropy, we propose a hierarchical pooling approach, SEP, to tackle the two issues. Specifically, without assigning the layer-specific compression quota, a global optimization algorithm is designed to generate the cluster assignment matrices for pooling at once. Then, we present an illustration of the local structure damage from previous methods in the reconstruction of ring and grid synthetic graphs. In addition to SEP, we further design two classification models, SEP-G and SEP-N for graph classification and node classification, respectively. The results show that SEP outperforms state-of-the-art graph pooling methods on graph classification benchmarks and obtains superior performance on node classifications.

In deep neural networks, better results can often be obtained by increasing the complexity of previously developed basic models. However, it is unclear whether there is a way to boost performance by decreasing the complexity of such models. Intuitively, given a problem, a simpler data structure comes with a simpler algorithm. Here, we investigate the feasibility of improving graph classification performance while simplifying the learning process. Inspired by structural entropy on graphs, we transform the data sample from graphs to coding trees, which is a simpler but essential structure for graph data. Furthermore, we propose a novel message passing scheme, termed hierarchical reporting, in which features are transferred from leaf nodes to root nodes by following the hierarchical structure of coding trees. We then present a tree kernel and a convolutional network to implement our scheme for graph classification. With the designed message passing scheme, the tree kernel and convolutional network have a lower runtime complexity of O(n) than Weisfeiler-Lehman subtree kernel and other graph neural networks of at least O(hm). We empirically validate our methods with several graph classification benchmarks and demonstrate that they achieve better performance and lower computational consumption than competing approaches.

We investigated a comprehensive analysis of the mutual exciting mechanism for the dynamic of stock price trends. A multi-dimensional Hawkes-model-based approach was proposed to capture the mutual exciting activities, which take the form of point processes induced by dual moving average crossovers. We first performed statistical measurements for the crossover event sequence, introducing the distribution of the inter-event times of dual moving average crossovers and the correlations of local variation (LV), which is often used in spike train analysis. It was demonstrated that the crossover dynamics in most stock sectors are generally more regular than a standard Poisson process, and the correlation between variations is ubiquitous. In this sense, the proposed model allowed us to identify some asymmetric cross-excitations, and a mutually exciting structure of stock sectors could be characterized by mutual excitation correlations obtained from the kernel matrix of our model. Using simulations, we were able to substantiate that a burst of the dual moving average crossovers in one sector increases the intensity of burst both in the same sector (self-excitation) as well as in other sectors (cross-excitation), generating episodes of highly clustered burst across the market. Furthermore, based on our finding, an algorithmic pair trading strategy was developed and backtesting results on real market data showed that the mutual excitation mechanism might be profitable for stock trading.

The preference of fund managers on various stocks forms the inner structure of the capital market. Data mining for the preference of funds from financial big data has attracted significant attention. In this paper, we study the preference features of Chinese capital market through mutual fund holdings data from 2010 to 2019. Complex fund-stock network structures are constructed according to the intersection of mutual fund managers' holdings. Further, a sub-graph extracting and embedding technology is introduced to make a quantitatively description of the preference of funds. Based on these embedding results, general fund correlation network can be constructed. The structure characteristics are demonstrated to be strongly correlated with performance of the funds. Empirical evidence from the financial data verifies the effectiveness of the proposed method.

The problem of planning a path in the real complex environment remains a difficult challenge although path planning has been studied extensively in the context of autonomous indoor mobile robots. One of the main reasons is lacking semantic information of the robot's working space. In this paper, we introduce the notion of space accessibility into path planning research. Semantic mapping is utilized to identify and recognize room and hallway on an occupancy grid map. A region topological map is built based on physical connection relation between regions and the types of regions. A multi-strategy path planning combining grid and region topological maps is developed. Paths in hallways will be selected preferentially and different planning methods are used for creating paths in different kinds of regions. Experimental evaluation is conducted on dozens of test maps with indoor layouts. Compared with paths produced by other techniques, the paths generated by our method are more like human beings decisions.

Path planning plays an important role in autonomous mobile robot. A lot of methods have been proposed in several decades. However, the complex environment in the real world becomes more challenging for robots. This paper conducts an experiment on evaluating path planning methods under different scenes. Four basic space configurations corresponding to specific indoor scenes are proposed. At the same time, the measurements of paths in these scenes are designed according to their different requirements for generating path. Following the common study in this field, we select six representative methods from four path planning families. The experimental results show the strength and the weakness of each method. More robust path planning methods for complex environments are expected from the view of application.