William Shen - Academia.edu (original) (raw)

Papers by William Shen

Supplementary Figure from Mutant IDH Inhibits IFNγ–TET2 Signaling to Promote Immunoevasion and Tu... more Supplementary Figure from Mutant IDH Inhibits IFNγ–TET2 Signaling to Promote Immunoevasion and Tumor Maintenance in Cholangiocarcinoma

Supplementary table from Mutant IDH Inhibits IFNγ–TET2 Signaling to Promote Immunoevasion and Tum... more Supplementary table from Mutant IDH Inhibits IFNγ–TET2 Signaling to Promote Immunoevasion and Tumor Maintenance in Cholangiocarcinoma

Proceedings of the International Conference on Automated Planning and Scheduling

We present the first approach capable of learning domain-independent planning heuristics entirely... more We present the first approach capable of learning domain-independent planning heuristics entirely from scratch. The heuristics we learn map the hypergraph representation of the delete-relaxation of the planning problem at hand, to a cost estimate that approximates that of the least-cost path from the current state to the goal through the hypergraph. We generalise Graph Networks to obtain a new framework for learning over hypergraphs, which we specialise to learn planning heuristics by training over state/value pairs obtained from optimal cost plans. Our experiments show that the resulting architecture, STRIPS-HGNs, is capable of learning heuristics that are competitive with existing delete-relaxation heuristics including LM-cut. We show that the heuristics we learn are able to generalise across different problems and domains, including to domains that were not seen during training.

Proceedings of the International Symposium on Combinatorial Search

We examine techniques for combining generalized policies with search algorithms to exploit the st... more We examine techniques for combining generalized policies with search algorithms to exploit the strengths and overcome the weaknesses of each when solving probabilistic planning problems. The Action Schema Network (ASNet) is a recent contribution to planning that uses deep learning and neural networks to learn generalized policies for probabilistic planning problems. ASNets are well suited to problems where local knowledge of the environment can be exploited to improve performance, but may fail to generalize to problems they were not trained on. Monte-Carlo Tree Search (MCTS) is a forward-chaining state space search algorithm for optimal decision making which performs simulations to incrementally build a search tree and estimate the values of each state. Although MCTS can achieve state-of-the-art results when paired with domain-specific knowledge, without this knowledge, MCTS requires a large number of simulations in order to obtain reliable state-value estimates. By combining ASNets...

British Journal of Surgery, 2021

Background Recurrence following resection of oesophago-gastric adenocarcinoma (OGA) is frequent a... more Background Recurrence following resection of oesophago-gastric adenocarcinoma (OGA) is frequent and associated with poor outcomes. Predictors of site, timing and mechanisms driving recurrence is poorly defined, which limits the development of anti-metastatic agents. The aim of this study was to investigate the patterns and timing of recurrence following resection of OGA. Methods Retrospective review of a prospectively maintained resection database from the Glasgow Royal Infirmary oesophago-gastric unit of patients undergoing surgery for OGA. Primary outcomes were recurrence and cancer specific death following surgery. Recurrence patterns were defined as liver, lung, peritoneal, locoregional only and other distant groups. The latter is a heterogenous group that do not include any liver, lung, or peritoneal metastases. Results N = 635 patients were identified having undergone surgical resection of OGA. Of these, n = 262 developed confirmed recurrent disease. Liver metastases (n = 86,...

Cancer Discovery, 2021

Isocitrate dehydrogenase 1 mutations (mIDH1) are common in cholangiocarcinoma. (R)-2-hydroxygluta... more Isocitrate dehydrogenase 1 mutations (mIDH1) are common in cholangiocarcinoma. (R)-2-hydroxyglutarate generated by the mIDH1 enzyme inhibits multiple α-ketoglutarate–dependent enzymes, altering epigenetics and metabolism. Here, by developing mIDH1-driven genetically engineered mouse models, we show that mIDH1 supports cholangiocarcinoma tumor maintenance through an immunoevasion program centered on dual (R)-2-hydroxyglutarate–mediated mechanisms: suppression of CD8+ T-cell activity and tumor cell–autonomous inactivation of TET2 DNA demethylase. Pharmacologic mIDH1 inhibition stimulates CD8+ T-cell recruitment and interferon γ (IFNγ) expression and promotes TET2-dependent induction of IFNγ response genes in tumor cells. CD8+ T-cell depletion or tumor cell–specific ablation of TET2 or IFNγ receptor 1 causes treatment resistance. Whereas immune-checkpoint activation limits mIDH1 inhibitor efficacy, CTLA4 blockade overcomes immunosuppression, providing therapeutic synergy. The findings ...

2019 IEEE/CVF International Conference on Computer Vision (ICCV), 2019

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017

Abstract Devices and methods of entraining a substance within an airflow are disclosed. Condensat... more Abstract Devices and methods of entraining a substance within an airflow are disclosed. Condensation aerosol delivery devices and methods of consistently producing multiple doses of a substance, such as a drug, having high purity, high yield, characterized by a particle size distribution appropriate for pulmonary delivery, and which can be administered to a user in a single dose are also disclosed.

SPIE Proceedings, 2008

The impact of embedded substrate defects on end-of-line die yield has become significant for adva... more The impact of embedded substrate defects on end-of-line die yield has become significant for advanced process technology nodes. Quality control and grading of wafers intended for leading-edge devices thus require effective detection and identification of embedded ...

Journal of Pharmacology and Experimental Therapeutics, 2004

Smoking involves heating a drug to form a mixture of drug vapor and gaseous degradation products.... more Smoking involves heating a drug to form a mixture of drug vapor and gaseous degradation products. These gases subsequently cool and condense into aerosol particles that are inhaled. Here, we demonstrate rapid and reliable systemic delivery of pure pharmaceutical compounds without degradation products through a related process that also involves inhalation of thermally generated aerosol. Drug is coated as a thin film on a metallic substrate and vaporized by heating

IEEE Journal of Solid-State Circuits, 2014

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018