paper-with-me

홈 › Papers

BESS: Balanced Entity Sampling and Sharing for Large-Scale Knowledge Graph Completion

2022-11-22 · Alberto Cattaneo, Daniel Justus, Harry Mellor, Douglas Orr, Jerome Maloberti, Zhenying Liu, Thorin Farnsworth, Andrew Fitzgibbon, Blazej Banaszewski, Carlo Luschi

We present the award-winning submission to the WikiKG90Mv2 track of OGB-LSC@NeurIPS 2022. The task is link-prediction on the large-scale knowledge graph WikiKG90Mv2, consisting of 90M+ nodes and 600M+ edges. Our solution uses a diverse ensemble of $85$ Knowledge Graph Embedding models combining five different scoring functions (TransE, TransH, RotatE, DistMult, ComplEx) and two different loss functions (log-sigmoid, sampled softmax cross-entropy). Each individual model is trained in parallel on a Graphcore Bow Pod$_{16}$ using BESS (Balanced Entity Sampling and Sharing), a new distribution framework for KGE training and inference based on balanced collective communications between workers. Our final model achieves a validation MRR of 0.2922 and a test-challenge MRR of 0.2562, winning the first place in the competition. The code is publicly available at: https://github.com/graphcore/distributed-kge-poplar/tree/2022-ogb-submission.

📄 PDF Abstract BibTeX arXiv:2211.12281

Code (2)

graphcore/distributed-kge-poplar 공식 구현 pytorch
graphcore-research/bess-kge pytorch

Tasks

Graph EmbeddingKnowledge Graph CompletionKnowledge Graph EmbeddingLink Prediction

Methods 이 논문이 사용한 방법론

Self-Adversarial Negative Sampling 설명 없음
RotatE RotatE is a method for generating graph embeddings which is able to model and infer various relation patterns including: symmetry/antisymmetry, inversion, and composition.…
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Similar Papers 제목 키워드 기반

Optimal Power Management for Large-Scale Battery Energy Storage Systems via Bayesian Inference

2025-03-04 · Amir Farakhor, Iman Askari, Di wu, Yebin Wang 외

Large-scale battery energy storage systems (BESS) have found ever-increasing use across industry and society to accelerate clean energy transition and improve energy supply reliability and resilience. However, their opti…

Bayesian InferenceManagementModel Predictive Controlparameter estimation

Physics-Aware Heterogeneous GNN Architecture for Real-Time BESS Optimization in Unbalanced Distribution Systems

2025-12-10 · Aoxiang Ma, Salah Ghamizi, Jun Cao, Pedro Rodriguez arxiv

Battery energy storage systems (BESS) have become increasingly vital in three-phase unbalanced distribution grids for maintaining voltage stability and enabling optimal dispatch. However, existing deep learning approache…

Efficient 3D affinely equivariant CNNs with adaptive fusion of augmented spherical Fourier-Bessel bases

2024-02-26 · Wenzhao Zhao, Steffen Albert, Barbara D. Wichtmann, Angelika Maurer 외

Filter-decomposition-based group equivariant convolutional neural networks (CNNs) have shown promising stability and data efficiency for 3D image feature extraction. However, these networks, which rely on parameter shari…

Image SegmentationMedical Image SegmentationOrgan SegmentationSegmentation+1

Optimization and Characterization of Near-Field Beams with Uniform Linear Arrays

2025-03-18 · Sota Uchimura, Josep Miquel Jornet, Koji Ishibashi

In this paper, we consider near-field beams that can mitigate signal attenuation and blockage effects using a uniform linear array (ULA). In particular, closed-form expressions for phase distributions in a ULA are derive…

Guaranteeing Service in Connected Microgrids: Storage Planning and Optimal Power Sharing Policy

2024-03-12 · Arnab Dey, Vivek Khatana, Ankur Mani, Murti V. Salapaka

The integration of renewable energy sources (RES) into power distribution grids poses challenges to system reliability due to the inherent uncertainty in their power production. To address this issue, battery energy sour…