Causal reasoning ability is crucial for numerous NLP applications. Despite the impressive emerging of ChatGPT in various tasks, it unclear how well performs causal reasoning. In this paper, we conduct first comprehensive evaluation ChatGPT's capabilities. Experiments show that not a good reasoner, but interpreter. Besides, has serious hallucination on reasoning, possibly due to reporting biases between and non-causal relationships natural language, as upgrading processes, such RLHF. The...

This paper investigates the problem of network embedding, which aims at learning low-dimensional vector representation nodes in networks. Most existing embedding methods rely solely on structure, i.e., linkage relationships between nodes, but ignore rich content information associated with it, is common real world networks and beneficial to describing characteristics a node. In this paper, we propose content-enhanced (CENE), capable jointly leveraging structure information. Our approach...

Recently, ChatGPT, a representative large language model (LLM), has gained considerable attention due to its powerful emergent abilities. Some researchers suggest that LLMs could potentially replace structured knowledge bases like graphs (KGs) and function as parameterized bases. However, while are proficient at learning probabilistic patterns based on corpus engaging in conversations with humans, they, previous smaller pre-trained models (PLMs), still have difficulty recalling facts...

Abstract Aiming at the problem of target position determination through multi-missile collaboration, algorithm research on estimation was carried out based line sight angle and missile distance measured by guidance head. Finally, “current” model uniform acceleration were used as motion models, an interactive multi-model filter constructed square root Cubature Kalman to achieve from serpentine maneuver motion. We have improved calculation method for transition probabilities various models in...

Large language models (LLMs) have shown great potential across various industries due to their remarkable ability generalize through instruction tuning. However, the limited availability of domain-specific data significantly hampers performance on specialized tasks. While existing methods primarily focus selecting training from general datasets that are similar target domain, they often fail consider joint distribution instructions, resulting in inefficient learning and suboptimal knowledge...

Inflammatory myofibroblastic tumors are extremely rare in the neck of infants, and pathological diagnosis may be challenging. Kinase fusions play an important role biology many inflammatory tumors, becoming effective diagnostic method. In this report, we present case East Asian (Han Chinese) patient with infant myofibroblastoma. DNA-based but not RNA-based next-generation sequencing was used to identify its targetable ROS1 fusions. This highlights importance simultaneously detecting DNA RNA...

Protection of direct current (DC) transmission lines is one the key difficulties to be urgently solved in construction future voltage-sourced converter (VSC)-based DC grids. In this paper, a novel ultra-high-speed traveling-wave (TW) protection principle for proposed which based on characteristics modulus voltage TWs. First, absolute value change amplitude 1-mode TW used construct starting-up element. Then, dyadic wavelet transform utilized extract wavelet-transform maxima (WTMM) and 0-mode...

Understanding causality has vital importance for various Natural Language Processing (NLP) applications. Beyond the labeled instances, conceptual explanations of can provide deep understanding causal fact to facilitate reasoning process. However, such explanation information still remains absent in existing resources. In this paper, we fill gap by presenting a human-annotated explainable CAusal REasoning dataset (e-CARE), which contains over 20K questions, together with natural language...

Existing methods regarding the influential nodes identification in complex networks usually assume that structures of are fully known. However, many cases, knowing full structure one network is hard or impossible, and each participant can only obtain partial networks. Therefore, be viewed as a private (PN) original network, they form multiple PNs. Then, question arises: how to collaboratively identify PNs while protecting privacy PN. To this end, secure multiparty computation ranking...

Network alignment aims to identify the corresponding nodes belonging same entity across different networks, which is a fundamental task in various applications. Existing embedding-based approaches usually involve two stages, namely embedding and matching. The stage conducts network on each capture primary structural regularity. In matching stage, mapping function built project learned embeddings latent space. However, these typically encounter challenges: (1) difficulty of unifying elusive...

Large Language Models (LLMs) have shown impressive capabilities in various applications, but they still face inconsistency issues. Existing works primarily focus on the issues within a single LLM, while we complementarily explore inter-consistency among multiple LLMs for collaboration. To examine whether can collaborate effectively to achieve consensus shared goal, commonsense reasoning, and introduce formal debate framework (FORD) conduct three-stage with real-world scenarios alignment:...

Graph neural networks (GNNs) are efficient techniques for learning graph representations and have shown remarkable success in tackling diverse graph-related tasks. However, the context of neighborhood aggregation paradigm, conventional GNNs limited capabilities capturing higher order structures topological semantics graphs. Researchers attempted to overcome this limitation by designing new that explore impacts motifs capture potentially information. existing motif-based often ignore lower...

Recently, the problem of user identification across multiple social networks (UIAMSNs) has attracted considerable attention since it is a prerequisite for many downstream tasks and applications. Although substantial network feature-based approaches have been proposed to solve UIAMSNs' problem, matching degree in most current works given by experience, which lacks solid theoretical basis. To alleviate above predicament, we propose algorithm based on naive Bayes model (UI-NBM) within...

To investigate the optimal management of patients with epidermal growth factor receptor gene (EGFR) mutant locally advanced non-small cell lung cancer (LA-NSCLC). Patients unresectable stage III adenocarcinoma (LAC) harboring EGFR mutations from 2012 to 2018 were analyzed retrospectively, and categorized into three groups according primary treatment: chemoradiotherpy (CRT) (group 1), combined radiation therapy (RT) EGFR-tyrosine kinase inhibitors (TKI) with/without chemotherapy 2), EGFR-TKI...

Many link prediction algorithms regarding single-layer social networks have been proposed, and however, how to predict interlayer links in multiplex is still the initial stage. In fact, of great significance, which closely related network security, product recommendation, mining, so forth. Given that many are sparse number first-order common matched neighbors (CMNs) very few, it not sufficient implement only based on CMNs. Moreover, scale-free property, leading roles CMNs significantly...

Link prediction is a fundamental task in network analysis, with the objective of predicting missing or potential links. While existing studies have mainly concentrated on single networks, it worth noting that numerous real-world networks exhibit interconnectedness. For example, individuals often register various social media platforms to access diverse services, such as chatting, tweeting, blogging, and rating movies. These share subset users are termed multilayer networks. The interlayer...

Reconstructing complex networks and predicting the dynamics are particularly challenging in real-world applications because available information data incomplete. We develop a unified collaborative deep-learning framework consisting of three modules: network inference, state estimation, dynamical learning. The complete structure is first inferred states unobserved nodes estimated, based on which learning module activated to determine evolution rules. An alternating parameter updating...

Due to the continuous technological innovation in industrial processes, many deep learning based methods have shown powerful capability handing equipment status monitoring, but most of them ignore temporal features and dynamic changes diverse spatial structure raw data. Meanwhile, these usually focus on handling a single downstream task, rarely consider different tasks simultaneously. To solve issues, this paper proposes more flexible monitoring framework dynamic-multilayer graph convolution...

Large-scale datasets in the real world inevitably involve label noise. Deep models can gradually overfit noisy labels and thus degrade model generalization. To mitigate effects of noise, learning with (LNL) methods are designed to achieve better generalization performance. Due lack suitable datasets, previous studies have frequently employed synthetic noise mimic real-world However, is not instance-dependent, making this approximation always effective practice. Recent research has proposed...