BiRank: Towards Ranking on Bipartite Graphs
BiRank: Towards Ranking on Bipartite Graphs(擴充套件到TriRank)
-
2017,IEEE,高明,何向南
-
解決問題:ranking vertices of a bipartite graph, graph ranking, bipartite
graphs對圖中的頂點進行排序. -
傳統方法問題:PageRank, HITS 只使用了graph structure
-
本文方法:graph link structure(就是定義的邊的權重,一個只與時間有關的單調、遞減、指數衰減函式), prior
information of vertices(query vector,其實就是用的頂點的鄰居數), regularization
function, fast convergence, (linear in the number of graph edges,
對稱歸一化。時間複雜度O(E) -
BiRank的輸入和輸出模型:1. 初始化:為使得模型快速收斂,p,u的初始化為STS和SST的主正交向量。
-
對比方法:View Count, Comment Count in the Past, Multivariate Linear model, PageRank, Co-HITS, BGER. 2. recommendation: Popularity, ItemKNN, PureSVD, PageRank, TagRW
-
資料集:Synthetic Random Graphs (邊服從uniform
distribution,10K*20K大小),Synthetic Power-law Graphs(生成圖的演算法見R-energy
for evaluating robustness of dynamic networks),Yelp和Amazon。 -
原始碼、評估指標:NDCG@K
-
實驗分析了哪些方面:模擬資料:1.收斂問題,迭代方法能否收斂到正則框架推理下的穩定值,迭代速度10輪;2:控制網路結構和查詢向量參與度的引數,與收斂速率的關係,查詢向量參與度越大,收斂越快;3.驗證一下時間複雜度線性於邊數量。真實資料第一個任務BiRank:1. 評估指標(Spearman coefficient),2. 引數調整(grid search,10% held out),3:各個演算法,評估指標對比,詳細的實驗分析(總體分析,拆開分析)。真實資料第二個任務TriRank:1.評估指標(NDCG@K)。
相關文章
- Hidden Bipartite Graph
- Graphs in PythonPython
- Dependencies for Graphs 閱讀筆記筆記
- Sphinx Ranking Mode(排序模式) (翻譯)排序模式
- Keys for graphs閱讀筆記筆記
- Cacti /graphs_new.php SQL Injection VulnerabilityPHPSQL
- Github Ranking:GitHub 全球 Developers, Organizations and Repositories 排行榜GithubDeveloper
- 巧用SQL Server(Ranking)實現view的排序功能SQLServerView排序
- 論文閱讀 Inductive Representation Learning on Temporal Graphs
- 論文解讀(BGRL)《Bootstrapped Representation Learning on Graphs》bootAPP
- 論文解讀(DAGNN)《Towards Deeper Graph Neural Networks》GNN
- 【backdoor attack】 POISONED FORGERY FACE: TOWARDS BACKDOOR ATTACKS ON FACE FORGERY DETECTION
- Sigir2024 ranking相關論文速讀
- 論文解讀《The Emerging Field of Signal Processing on Graphs》
- 論文閱讀 TEMPORAL GRAPH NETWORKS FOR DEEP LEARNING ON DYNAMIC GRAPHS
- 閱讀筆記-MoFlow: An Invertible Flow Model for Generating Molecular Graphs筆記
- 論文解讀(GROC)《Towards Robust Graph Contrastive Learning》AST
- 論文解讀(AutoSSL)《Automated Self-Supervised Learning for Graphs》
- Smooth-AP: Smoothing the Path Towards Large-Scale Image Retrieval(翻譯)
- 論文解讀(GraphSMOTE)《GraphSMOTE: Imbalanced Node Classification on Graphs with Graph Neural Networks》
- 論文解讀(MVGRL)Contrastive Multi-View Representation Learning on GraphsASTView
- 《Towards Good Practices for Very Deep Two-Stream ConvNets》閱讀筆記Go筆記
- 2018ACM-ICPC北京賽區 - A:Jin Yong’s Wukong Ranking List(DFS)ACM
- 論文解讀(MGAE)《MGAE: Masked Autoencoders for Self-Supervised Learning on Graphs》
- 論文解讀(ClusterSCL)《ClusterSCL: Cluster-Aware Supervised Contrastive Learning on Graphs》AST
- 論文解讀(ValidUtil)《Rethinking the Setting of Semi-supervised Learning on Graphs》Thinking
- 論文解讀(USIB)《Towards Explanation for Unsupervised Graph-Level Representation Learning》
- 論文解讀(DCN)《Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering》
- 論文翻譯:2021_Towards model compression for deep learning based speech enhancement
- [Paper Reading] GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion ModelsIDEGUI
- BCN Ranking:2021年Intel在日本DIY市場反殺AMD 市場份額達74%Intel
- 論文解讀二代GCN《Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering》GCASTZedFilter
- Connecting the Dots: Document-level Neural Relation Extraction with Edge-oriented Graphs 關係抽取論文總結
- [論文][表情識別]Towards Semi-Supervised Deep Facial Expression Recognition with An Adaptive Confidence MarginExpressAPTIDE
- 深度學習論文翻譯解析(十三):Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks深度學習ASTCNNObject
- BCN Ranking:2020年5月AMD第三代處理器銷售份額佔比高達67.4%