arXiv subjects

Zhen Yao

Publications and source records attributed to Zhen Yao.

31 records · Page 2Linked to original sources

arXiv2023

Target-oriented Sentiment Classification with Sequential Cross-modal Semantic Graph

Multi-modal aspect-based sentiment classification (MABSC) is task of classifying the sentiment of a target entity mentioned in a sentence and an image. However, previous methods failed to account for the fine-grained semantic association between the image and the text, which resulted in limited identification of fine-grained image aspects and opinions. To address these limitations, in this paper we propose a new approach called SeqCSG, which enhances the encoder-decoder sentiment classification framework using sequential cross-modal semantic graphs. SeqCSG utilizes image captions and scene graphs to extract both global and local fine-grained image information and considers them as elements of the cross-modal semantic graph along with tokens from tweets. The sequential cross-modal semantic graph is represented as a sequence with a multi-modal adjacency matrix indicating relationships between elements. Experimental results show that the approach outperforms existing methods and achieves state-of-the-art performance on two standard datasets. Further analysis has demonstrated that the model can implicitly learn the correlation between fine-grained information of the image and the text with the given target. Our code is available at https://github.com/zjukg/SeqCSG.

cs.CV↗

arXiv2023

NeuralKG-ind: A Python Library for Inductive Knowledge Graph Representation Learning

Since the dynamic characteristics of knowledge graphs, many inductive knowledge graph representation learning (KGRL) works have been proposed in recent years, focusing on enabling prediction over new entities. NeuralKG-ind is the first library of inductive KGRL as an important update of NeuralKG library. It includes standardized processes, rich existing methods, decoupled modules, and comprehensive evaluation metrics. With NeuralKG-ind, it is easy for researchers and engineers to reproduce, redevelop, and compare inductive KGRL methods. The library, experimental methodologies, and model re-implementing results of NeuralKG-ind are all publicly released at https://github.com/zjukg/NeuralKG/tree/ind .

cs.AI↗

arXiv2023

Tele-Knowledge Pre-training for Fault Analysis

In this work, we share our experience on tele-knowledge pre-training for fault analysis, a crucial task in telecommunication applications that requires a wide range of knowledge normally found in both machine log data and product documents. To organize this knowledge from experts uniformly, we propose to create a Tele-KG (tele-knowledge graph). Using this valuable data, we further propose a tele-domain language pre-training model TeleBERT and its knowledge-enhanced version, a tele-knowledge re-training model KTeleBERT. which includes effective prompt hints, adaptive numerical data encoding, and two knowledge injection paradigms. Concretely, our proposal includes two stages: first, pre-training TeleBERT on 20 million tele-related corpora, and then re-training it on 1 million causal and machine-related corpora to obtain KTeleBERT. Our evaluation on multiple tasks related to fault analysis in tele-applications, including root-cause analysis, event association prediction, and fault chain tracing, shows that pre-training a language model with tele-domain data is beneficial for downstream tasks. Moreover, the KTeleBERT re-training further improves the performance of task models, highlighting the effectiveness of incorporating diverse tele-knowledge into the model.

cs.AI↗

arXiv2023

Analogical Inference Enhanced Knowledge Graph Embedding

Knowledge graph embedding (KGE), which maps entities and relations in a knowledge graph into continuous vector spaces, has achieved great success in predicting missing links in knowledge graphs. However, knowledge graphs often contain incomplete triples that are difficult to inductively infer by KGEs. To address this challenge, we resort to analogical inference and propose a novel and general self-supervised framework AnKGE to enhance KGE models with analogical inference capability. We propose an analogical object retriever that retrieves appropriate analogical objects from entity-level, relation-level, and triple-level. And in AnKGE, we train an analogy function for each level of analogical inference with the original element embedding from a well-trained KGE model as input, which outputs the analogical object embedding. In order to combine inductive inference capability from the original KGE model and analogical inference capability enhanced by AnKGE, we interpolate the analogy score with the base model score and introduce the adaptive weights in the score function for prediction. Through extensive experiments on FB15k-237 and WN18RR datasets, we show that AnKGE achieves competitive results on link prediction task and well performs analogical inference.

cs.AI↗

arXiv2023

Entity-Agnostic Representation Learning for Parameter-Efficient Knowledge Graph Embedding

We propose an entity-agnostic representation learning method for handling the problem of inefficient parameter storage costs brought by embedding knowledge graphs. Conventional knowledge graph embedding methods map elements in a knowledge graph, including entities and relations, into continuous vector spaces by assigning them one or multiple specific embeddings (i.e., vector representations). Thus the number of embedding parameters increases linearly as the growth of knowledge graphs. In our proposed model, Entity-Agnostic Representation Learning (EARL), we only learn the embeddings for a small set of entities and refer to them as reserved entities. To obtain the embeddings for the full set of entities, we encode their distinguishable information from their connected relations, k-nearest reserved entities, and multi-hop neighbors. We learn universal and entity-agnostic encoders for transforming distinguishable information into entity embeddings. This approach allows our proposed EARL to have a static, efficient, and lower parameter count than conventional knowledge graph embedding methods. Experimental results show that EARL uses fewer parameters and performs better on link prediction tasks than baselines, reflecting its parameter efficiency.

cs.CL↗

arXiv2022

Meta-Learning Based Knowledge Extrapolation for Knowledge Graphs in the Federated Setting

We study the knowledge extrapolation problem to embed new components (i.e., entities and relations) that come with emerging knowledge graphs (KGs) in the federated setting. In this problem, a model trained on an existing KG needs to embed an emerging KG with unseen entities and relations. To solve this problem, we introduce the meta-learning setting, where a set of tasks are sampled on the existing KG to mimic the link prediction task on the emerging KG. Based on sampled tasks, we meta-train a graph neural network framework that can construct features for unseen components based on structural information and output embeddings for them. Experimental results show that our proposed method can effectively embed unseen components and outperforms models that consider inductive settings for KGs and baselines that directly use conventional KG embedding methods.

cs.CL↗

arXiv2022

NeuralKG: An Open Source Library for Diverse Representation Learning of Knowledge Graphs

NeuralKG is an open-source Python-based library for diverse representation learning of knowledge graphs. It implements three different series of Knowledge Graph Embedding (KGE) methods, including conventional KGEs, GNN-based KGEs, and Rule-based KGEs. With a unified framework, NeuralKG successfully reproduces link prediction results of these methods on benchmarks, freeing users from the laborious task of reimplementing them, especially for some methods originally written in non-python programming languages. Besides, NeuralKG is highly configurable and extensible. It provides various decoupled modules that can be mixed and adapted to each other. Thus with NeuralKG, developers and researchers can quickly implement their own designed models and obtain the optimal training methods to achieve the best performance efficiently. We built an website in http://neuralkg.zjukg.cn to organize an open and shared KG representation learning community. The source code is all publicly released at https://github.com/zjukg/NeuralKG.

cs.LG↗

arXiv2013

Thermal Conductivity and Phonon Transport in Suspended Few-Layer Hexagonal Boron Nitride

The thermal conductivity of suspended few-layer hexagonal boron nitride (h-BN) was measured using a micro-bridge device with built-in resistance thermometers. Based on the measured thermal resistance values of 11-12 atomic layer h-BN samples with suspended length ranging between 3 and 7.5 um, the room-temperature thermal conductivity of a 11-layer sample was found to be about 360 Wm-1K-1, approaching the basal plane value reported for bulk h-BN. The presence of a polymer residue layer on the sample surface was found to decrease the thermal conductivity of a 5-layer h-BN sample to be about 250 Wm-1K-1 at 300 K. Thermal conductivities for both the 5 layer and the 11 layer samples are suppressed at low temperatures, suggesting increasing scattering of low frequency phonons in thin h-BN samples by polymer residue.

cond-mat.mtrl-sci↗

arXiv2009

Realization of a High Mobility Dual-gated Graphene Field Effect Transistor with Al2O3 Dielectric

We fabricate and characterize dual-gated graphene field-effect transistors (FETs) using Al2O3 as top-gate dielectric. We use a thin Al film as a nucleation layer to enable the atomic layer deposition of Al2O3. Our devices show mobility values of over 8,000 cm2/Vs at room temperature, a finding which indicates that the top-gate stack does not significantly increase the carrier scattering, and consequently degrade the device characteristics. We propose a device model to fit the experimental data using a single mobility value.

cond-mat.mes-hall↗

arXiv2002

Fabrication of nanometer-spaced electrodes using gold nanoparticles

A simple and highly reproducible technique is demonstrated for the fabrication of metallic electrodes with nanometer separation. Commercially available bare gold colloidal nanoparticles are first trapped between prefabricated large-separation electrodes to form a low-resistance bridge by an ac electric field. A large dc voltage is then applied to break the bridge via electromigration at room temperature, which consistently produces gaps in the sub-10 nm range. The technique is readily applied to prefabricated electrodes with separation up to 1 micron, which can be defined using optical lithography. The simple fabrication scheme will facilitate electronic transport studies of individual nanostructures made by chemical synthesis. As an example, measurement of a thiol-coated gold nanoparticle showing a clear Coulomb staircase is presented.

cond-mat.mes-hall↗

arXiv2000

Electrical transport through carbon nanotube junctions created by mechanical manipulation

Using an atomic force microscope we have created nanotube junctions such as buckles and crossings within individual single-wall metallic carbon nanotubes connected to metallic electrodes. The electronic transport properties of these manipulated structures show that they form electronic tunnel junctions. The conductance shows power-law behavior as a function of bias voltage and temperature, which can be well modeled by a Luttinger liquid model for tunneling between two nanotube segments separated by the manipulated junction.

cond-mat.mes-hall↗

arXiv1999

High-Field Electrical Transport in Single-Wall Carbon Nanotubes

Using low-resistance electrical contacts, we have measured the intrinsic high-field transport properties of metallic single-wall carbon nanotubes. Individual nanotubes appear to be able to carry currents with a density exceeding 10^9 A/cm^2. As the bias voltage is increased, the conductance drops dramatically due to scattering of electrons. We show that the current-voltage characteristics can be explained by considering optical or zone-boundary phonon emission as the dominant scattering mechanism at high field.

cond-mat.mes-hall↗

arXiv1999

Structure of Flux Line Lattices with Weak Disorder at Large Length Scales

Dislocation-free decoration images containing up to 80,000 vortices have been obtained on high quality Bi$_{2}$Sr$_{2}$CaCu$_{2}$O$_{8+x}$ superconducting single crystals. The observed flux line lattices are in the random manifold regime with a roughening exponent of 0.44 for length scales up to 80-100 lattice constants. At larger length scales, the data exhibit nonequilibrium features that persist for different cooling rates and field histories.

cond-mat.supr-con↗