Peking University

About: Peking University is a education organization based out in Beijing, Beijing, China. It is known for research contribution in the topics: Population & Cancer. The organization has 143902 authors who have published 181007 publications receiving 4103666 citations. The organization is also known as: PKU & Beida.

Erlotinib versus chemotherapy as first-line treatment for patients with advanced EGFR mutation-positive non-small-cell lung cancer (OPTIMAL, CTONG-0802): a multicentre, open-label, randomised, phase 3 study

Abstract: Summary Background Activating mutations in EGFR are important markers of response to tyrosine kinase inhibitor (TKI) therapy in non-small-cell lung cancer (NSCLC). The OPTIMAL study compared efficacy and tolerability of the TKI erlotinib versus standard chemotherapy in the first-line treatment of patients with advanced EGFR mutation-positive NSCLC. Methods We undertook an open-label, randomised, phase 3 trial at 22 centres in China. Patients older than 18 years with histologically confirmed stage IIIB or IV NSCLC and a confirmed activating mutation of EGFR (exon 19 deletion or exon 21 L858R point mutation) received either oral erlotinib (150 mg/day) until disease progression or unacceptable toxic effects, or up to four cycles of gemcitabine plus carboplatin. Patients were randomly assigned (1:1) with a minimisation procedure and were stratified according to EGFR mutation type, histological subtype (adenocarcinoma vs non-adenocarcinoma), and smoking status. The primary outcome was progression-free survival, analysed in patients with confirmed disease who received at least one dose of study treatment. The trial is registered at ClinicalTrials.gov, number NCT00874419, and has completed enrolment; patients are still in follow-up. Findings 83 patients were randomly assigned to receive erlotinib and 82 to receive gemcitabine plus carboplatin; 82 in the erlotinib group and 72 in the chemotherapy group were included in analysis of the primary endpoint. Median progression-free survival was significantly longer in erlotinib-treated patients than in those on chemotherapy (13.1 [95% CI 10.58–16.53] vs 4.6 [4.21–5.42] months; hazard ratio 0.16, 95% CI 0.10–0.26; p vs no patients with either event on erlotinib); the most common grade 3 or 4 toxic effects with erlotinib were increased alanine aminotransferase concentrations (three [4%] of 83 patients) and skin rash (two [2%] patients). Chemotherapy was also associated with increased treatment-related serious adverse events (ten [14%] of 72 patients [decreased platelet count, n=8; decreased neutrophil count, n=1; hepatic dysfunction, n=1] vs two [2%] of 83 patients [both hepatic dysfunction]). Interpretation Compared with standard chemotherapy, erlotinib conferred a significant progression-free survival benefit in patients with advanced EGFR mutation-positive NSCLC and was associated with more favourable tolerability. These findings suggest that erlotinib is important for first-line treatment of patients with advanced EGFR mutation-positive NSCLC. Funding F Hoffmann-La Roche Ltd (China); Science and Technology Commission of Shanghai Municipality.

LINE: Large-scale Information Network Embedding

Abstract: This paper studies the problem of embedding very large information networks into low-dimensional vector spaces, which is useful in many tasks such as visualization, node classification, and link prediction. Most existing graph embedding methods do not scale for real world information networks which usually contain millions of nodes. In this paper, we propose a novel network embedding method called the ``LINE,'' which is suitable for arbitrary types of information networks: undirected, directed, and/or weighted. The method optimizes a carefully designed objective function that preserves both the local and global network structures. An edge-sampling algorithm is proposed that addresses the limitation of the classical stochastic gradient descent and improves both the effectiveness and the efficiency of the inference. Empirical experiments prove the effectiveness of the LINE on a variety of real-world information networks, including language networks, social networks, and citation networks. The algorithm is very efficient, which is able to learn the embedding of a network with millions of vertices and billions of edges in a few hours on a typical single machine. The source code of the LINE is available online\footnote{\url{https://github.com/tangjianpku/LINE}}.

LINE: Large-scale Information Network Embedding

Abstract: This paper studies the problem of embedding very large information networks into low-dimensional vector spaces, which is useful in many tasks such as visualization, node classification, and link prediction. Most existing graph embedding methods do not scale for real world information networks which usually contain millions of nodes. In this paper, we propose a novel network embedding method called the "LINE," which is suitable for arbitrary types of information networks: undirected, directed, and/or weighted. The method optimizes a carefully designed objective function that preserves both the local and global network structures. An edge-sampling algorithm is proposed that addresses the limitation of the classical stochastic gradient descent and improves both the effectiveness and the efficiency of the inference. Empirical experiments prove the effectiveness of the LINE on a variety of real-world information networks, including language networks, social networks, and citation networks. The algorithm is very efficient, which is able to learn the embedding of a network with millions of vertices and billions of edges in a few hours on a typical single machine. The source code of the LINE is available online.

Face recognition using Laplacianfaces

Abstract: We propose an appearance-based face recognition method called the Laplacianface approach. By using locality preserving projections (LPP), the face images are mapped into a face subspace for analysis. Different from principal component analysis (PCA) and linear discriminant analysis (LDA) which effectively see only the Euclidean structure of face space, LPP finds an embedding that preserves local information, and obtains a face subspace that best detects the essential face manifold structure. The Laplacianfaces are the optimal linear approximations to the eigenfunctions of the Laplace Beltrami operator on the face manifold. In this way, the unwanted variations resulting from changes in lighting, facial expression, and pose may be eliminated or reduced. Theoretical analysis shows that PCA, LDA, and LPP can be obtained from different graph models. We compare the proposed Laplacianface approach with Eigenface and Fisherface methods on three different face data sets. Experimental results suggest that the proposed Laplacianface approach provides a better representation and achieves lower error rates in face recognition.