Yufan Guo

Yufan Guo

Auto-generated profile, based on publicly available information

Yufan Guo

Language Technology Laboratory, TAL, University of Cambridge, 9 West Road, Cambridge, CB39DB, UK

Research Focus & Metrics

Academic Career Length

13Years (2014-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
17 January 2017 in BMC Bioinformatics
Verified
Journal Article
Open access
PDF file
YLYiding Lu
Yiding Lu
YGYufan Guo
Yufan Guo
AKAnna Korhonen
Anna Korhonen
In silico drug-target interaction (DTI) prediction plays an integral role in drug repositioning: the discovery of new uses for existing drugs. One popular method of drug repositioning is network-based DTI prediction, which uses complex network theory to predict DTIs from a drug-target network. Currently, most network-based DTI prediction is based on machine learning – methods such as Restricted Boltzmann Machines (RBM) or Support Vector Machines (SVM). These methods require additional information about the characteristics of drugs, targets and DTIs, such as chemical structure, genome sequence, binding types, causes of interactions, etc., and do not perform satisfactorily when such information is unavailable. We propose a new, alternative method for DTI prediction that makes use of only network topology information attempting to solve this problem. We compare our method for DTI prediction against the well-known RBM approach. We show that when applied to the MATADOR database, our approach based on node neighborhoods yield higher precision for high-ranking predictions than RBM when no information regarding DTI types is available. This demonstrates that approaches purely based on network topology provide a more suitable approach to DTI prediction in the many real-life situations where little or no prior knowledge is available about the characteristics of drugs, targets, or their interactions.

21 May 2018 in BMC Bioinformatics
Verified
Journal Article
Open access
PDF file
GCGamal Crichton
Gamal Crichton
YGYufan Guo
Yufan Guo
SPSampo Pyysalo
Sampo Pyysalo
AKAnna Korhonen
Anna Korhonen
Link prediction in biomedical graphs has several important applications including predicting Drug-Target Interactions (DTI), Protein-Protein Interaction (PPI) prediction and Literature-Based Discovery (LBD). It can be done using a classifier to output the probability of link formation between nodes. Recently several works have used neural networks to create node representations which allow rich inputs to neural classifiers. Preliminary works were done on this and report promising results. However they did not use realistic settings like time-slicing, evaluate performances with comprehensive metrics or explain when or why neural network methods outperform. We investigated how inputs from four node representation algorithms affect performance of a neural link predictor on random- and time-sliced biomedical graphs of real-world sizes (∼ 6 million edges) containing information relevant to DTI, PPI and LBD. We compared the performance of the neural link predictor to those of established baselines and report performance across five metrics. In random- and time-sliced experiments when the neural network methods were able to learn good node representations and there was a negligible amount of disconnected nodes, those approaches outperformed the baselines. In the smallest graph (∼ 15,000 edges) and in larger graphs with approximately 14% disconnected nodes, baselines such as Common Neighbours proved a justifiable choice for link prediction. At low recall levels (∼ 0.3) the approaches were mostly equal, but at higher recall levels across all nodes and average performance at individual nodes, neural network approaches were superior. Analysis showed that neural network methods performed well on links between nodes with no previous common neighbours; potentially the most interesting links. Additionally, while neural network methods benefit from large amounts of data, they require considerable amounts of computational resources to utilise them. Our results indicate that when there is enough data for the neural network methods to use and there are a negligible amount of disconnected nodes, those approaches outperform the baselines. At low recall levels the approaches are mostly equal but at higher recall levels and average performance at individual nodes, neural network approaches are superior. Performance at nodes without common neighbours which indicate more unexpected and perhaps more useful links account for this.

1 September 2014 in Toxicology Letters
Verified
Journal Article
Closed access
ISIlona Silins
Ilona Silins
AKAnna Korhonen
Anna Korhonen
YGYufan Guo
Yufan Guo
USUlla Stenius
Ulla Stenius

1 September 2014 in Toxicology Letters
Verified
Journal Article
Closed access
KLKristin Larsson
Kristin Larsson
ISIlona Silins
Ilona Silins
YGYufan Guo
Yufan Guo
AKAnna Korhonen
Anna Korhonen
USUlla Stenius
Ulla Stenius
MBMarika Berglund
Marika Berglund

30 November 2017
Verified
Book
Closed access
TPThierry Poibeau
Thierry Poibeau
AVAline Villavicencio
Aline Villavicencio
PBPhilippe Blache
Philippe Blache
BMBrian Murphy
Brian Murphy
LWLeila Wehbe
Leila Wehbe
RTRemi Van Trijp
Remi Van Trijp
How do infants learn a language? Why and how do languages evolve? How do we understand a sentence? This book explores these questions using recent computational models that shed new light on issues related to language and cognition. The chapters in this collection propose original analyses of specific problems and develop computational models that have been tested and evaluated on real data. Featuring contributions from a diverse group of experts, this interdisciplinary book bridges the gap between natural language processing and cognitive sciences. It is divided into three sections, focusing respectively on models of neural and cognitive processing, data driven methods, and social issues in language evolution. This book will be useful to any researcher and advanced student interested in the analysis of the links between the brain and the language faculty.

30 November 2017
Verified
Book Chapter
Closed access
XJXiao Jiang
Xiao Jiang
YHYan Huang
Yan Huang
YGYufan Guo
Yufan Guo
JGJeroen Geertzen
Jeroen Geertzen
TATheodora Alexopoulou
Theodora Alexopoulou
AKAnna Korhonen
Anna Korhonen
Native Language Identification (NLI) is a task aimed at determining the native language (L1) of learners of second language (L2) on the basis of their written texts. To date, research on NLI has focused on relatively small corpora. We apply NLI to EFCAMDAT, an L2 English learner corpus that is not only multiple times larger than previous L2 corpora but also provides pseudo-longitudinal data across several proficiency levels. Based on accurate machine learning with a wide range of linguistic features, our investigation reveals interesting patterns in the longitudinal data that are useful for both further development of NLI and its application to research on L2 acquisition. Introduction Native language identification (NLI) is a task aimed at detecting the native language (L1) of writers on the basis of their second language (L2) production. NLI is important for natural language processing (NLP) applications including language tutoring systems and authorship profiling. Moreover, NLI can offer useful empirical data for research on L2 acquisition. For example, NLI can shed light on how L1 background influences L2 learning, and on differences between the writings of L2 learners across different L1 backgrounds. To date, studies on NLI have focused on relatively small learner corpora. Furthermore, none of them have investigated the influence of L1s across L2 proficiency levels. Our work takes the first step toward addressing these problems. We apply NLI to EFCAMDAT, the EF-Cambridge Open Language Database (Geertzen, Alexopoulou, and Korhonen, 2013), an open-access L2 learner corpus. EFCAMDAT consists of writings of learners submitted to Englishtown, the online school of EF. EFCAMDAT stands out for its size, diversity of student backgrounds, and coverage of the proficiency levels. The first release of 2013 (Geertzen, Alexopoulou, and Korhonen, 2013), on which this paper is based, amounts to 30 million words, a corpus multiple times larger than any other available L2 corpora. Using a standard machine learning–based methodology for NLI, we explore the optimal linguistic features for NLI on this data at different proficiency levels. We discover interesting patterns that can be useful for both further development of NLI and its application to research on L2 acquisition. In this introductory section, we first review the history of research on NLI, and introduce the data sets that have been used in earlier NLI research.We then summarise our contribution briefly.