Graph similarity search is among the most important graph-based applications, e.g. finding chemical compounds that are similar to a query compound. similarity/distance computation, such as Edit Distance (GED) and Maximum Common Subgraph (MCS), core operation of graph many other but very costly compute in practice. Inspired by recent success neural network approaches several node or classification, we propose novel based approach address this classic yet challenging problem, aiming alleviate...

Graph similarity computation is one of the core operations in many graph-based applications, such as graph search, database analysis, clustering, etc. Since computing exact distance/similarity between two graphs typically NP-hard, a series approximate methods have been proposed with trade-off accuracy and speed. Recently, several data-driven approaches based on neural networks proposed, most which model graph-graph inner product their graph-level representations, different techniques for...

We introduce a novel approach to graph-level representation learning, which is embed an entire graph into vector space where the embeddings of two graphs preserve their graph-graph proximity. Our approach, UGraphEmb, general framework that provides means performing embedding in completely unsupervised and inductive manner. The learned neural network can be considered as function receives any input, either seen or unseen training set, transforms it embedding. A generation mechanism called...

Using High-Level Synthesis (HLS), the hardware designers must describe only a high-level behavioral flow of design. However, it still can take weeks to develop high-performance architecture mainly because there are many design choices at higher level explore. Besides, takes several minutes hours evaluate with HLS tool. To solve this problem, we model tool graph neural network that is trained be used for wide range applications. The experimental results demonstrate our estimate quality in...

Hardware design presents numerous challenges stemming from its complexity and advancing technologies. These result in longer turn-around-time (TAT) for optimizing performance, power, area, cost (PPAC) during synthesis, verification, physical design, reliability loops. Large Language Models (LLMs) have shown remarkable capacity to comprehend generate natural language at a massive scale, leading many potential applications benefits across various domains. Successful LLM-based agents hardware...

Graph similarity search is among the most important graph-based applications, e.g. finding chemical compounds that are similar to a query compound. computation, such as Edit Distance (GED) and Maximum Common Subgraph (MCS), core operation of graph many other but very costly compute in practice. Inspired by recent success neural network approaches several node or classification, we propose novel based approach address this classic yet challenging problem, aiming alleviate computational burden...

Graph similarity search aims to find the most similar graphs a query in graph database terms of given proximity measure, say Edit Distance (GED). It is widely studied yet still challenging problem. Most studies are based on pruning-verification framework, which first prunes non-promising and then conducts verification small candidate set. Existing methods capable managing databases with thousands or tens graphs, but fail scale even larger database, due their exact pruning strategy. Inspired...

Open Source Software (OSS) is forming the spines of technology infrastructures, attracting millions talents to contribute. Notably, it challenging and critical consider both developers' interests semantic features project code recommend appropriate development tasks OSS developers. In this paper, we formulate novel problem recommendation, whose purpose predict future contribution behaviors developers given their interaction history, source code, hierarchical file structures projects....

The efficient and timely optimization of microarchitecture for a target application is hindered by the long evaluation runtime design candidate, creating serious burden. To tackle this problem, researchers have started using learning algorithms such as graph neural networks (GNNs) to accelerate process developing surrogate tool. However, challenges arise when models HLS tools due program's dependency range deeply coupled input program transformations (i.e., pragmas). address them, in paper,...

We introduce GSimCNN (Graph Similarity Computation via Convolutional Neural Networks) for predicting the similarity score between two graphs. As core operation of graph search, pairwise computation is a challenging problem due to NP-hard nature computing many distance/similarity metrics. demonstrate our model using Graph Edit Distance (GED) as example metric. Experiments on three real datasets that achieves state-of-the-art performance search.

Two-view knowledge graphs (KGs) jointly represent two components: an ontology view for abstract and commonsense concepts, instance specific entities that are instantiated from ontological concepts. As such, these KGs contain heterogeneous structures hierarchical, the ontology-view, cyclical, instance-view. Despite various in KGs, recent works on embedding assume entire KG belongs to only one of views but not both simultaneously. For seek put together, assumed belong same geometric space,...

We introduce a novel approach to graph-level representation learning, which is embed an entire graph into vector space where the embeddings of two graphs preserve their graph-graph proximity. Our approach, UGRAPHEMB, general framework that provides means performing embedding in completely unsupervised and inductive manner. The learned neural network can be considered as function receives any input, either seen or unseen training set, transforms it embedding. A generation mechanism called...

Experimental observations from different fiber lasers or states within a laser confirm that the dispersion parameter is wavelength-dependent, which suggests measurement based on Kelly sidebands should base either positive negative sideband order only rather than orders across zero. © 2016 Wiley Periodicals, Inc. Microwave Opt Technol Lett 58:242–245,

Abstract We have developed ACROBAT (Annotation for Case Reports using Open Biomedical Annotation Terms), a typing system detailed information extraction from clinical text. This resource supports identification and categorization of entities, events, relations within text documents, including clincal case reports (CCRs) the free-text components electronic health records. Using 200 CCRs, we annotated wide variety real-world disease presentations. The resulting dataset, MACCROBAT2018, is rich...

We introduce Bi-GNN for modeling biological link prediction tasks such as drug-drug interaction (DDI) and protein-protein (PPI). Taking an example, existing methods using machine learning either only utilize the structure between drugs without graph representation of each drug molecule, or leverage individual compound structures higher-level DDI graph. The key idea our method is to fundamentally view data a bi-level graph, where highest level represents entities (interaction graph), entity...

Recently, there is a growing interest in developing learning-based models as surrogate of the High-Level Synthesis (HLS) tools, where key objective rapid prediction quality candidate HLS design for automated space exploration (DSE). Training usually conducted on given set computation kernels (or short) needed hardware acceleration. However, model must also perform well new kernels. The discrepancy between training and kernels, called domain shift, frequently leads to accuracy drop which turn...

High-level synthesis (HLS) has freed the computer architects from developing their designs in a very low-level language and needing to exactly specify how data should be transferred register-level. With help of HLS, hardware designers must describe only high-level behavioral flow design. Despite this, it still can take weeks develop high-performance architecture mainly because there are many design choices at higher level that requires more time explore. It also takes several minutes hours...

In recent years, domain-specific accelerators (DSAs) have gained popularity for applications such as deep learning and autonomous driving. To facilitate DSA designs, programmers use high-level synthesis (HLS) to compile a description written in C/C++ into design with low-level hardware languages that eventually synthesize DSAs on circuits. However, creating high-quality HLS still demands significant domain knowledge, particularly microarchitecture decisions expressed \textit{pragmas}. Thus,...

High-level synthesis (HLS) is an automated design process that transforms high-level code into hardware designs, enabling the rapid development of accelerators. HLS relies on pragmas, which are directives inserted source to guide process, and pragmas have various settings values significantly impact resulting design. State-of-the-art ML-based methods, such as HARP, first train a deep learning model, typically based graph neural networks (GNNs) applied graph-based representations pragmas....