Clare Zhu - Academia.edu (original) (raw)

Papers by Clare Zhu

Research paper thumbnail of Effective Approaches to Batch Parallelization for Dynamic Neural Network Architectures

arXiv (Cornell University), Jul 8, 2017

We present a simple dynamic batching approach applicable to a large class of dynamic architecture... more We present a simple dynamic batching approach applicable to a large class of dynamic architectures that consistently yields speedups of over 10x. We provide performance bounds when the architecture is not known a priori and a stronger bound in the special case where the architecture is a predetermined balanced tree. We evaluate our approach on Johnson et al.'s recent visual question answering (VQA) result of his CLEVR dataset by Inferring and Executing Programs (IEP). We also evaluate on sparsely gated mixture of experts layers and achieve speedups of up to 1000x over the naive implementation.

Research paper thumbnail of CCDC 1489645: Experimental Crystal Structure Determination

An entry from the Cambridge Structural Database, the world's repository for small molecule cr... more An entry from the Cambridge Structural Database, the world's repository for small molecule crystal structures. The entry contains experimental data from a crystal diffraction study. The deposited dataset for this entry is freely available from the CCDC and typically includes 3D coordinates, cell parameters, space group, experimental conditions and quality measures.

Research paper thumbnail of CCDC 1489644: Experimental Crystal Structure Determination

An entry from the Cambridge Structural Database, the world's repository for small molecule cr... more An entry from the Cambridge Structural Database, the world's repository for small molecule crystal structures. The entry contains experimental data from a crystal diffraction study. The deposited dataset for this entry is freely available from the CCDC and typically includes 3D coordinates, cell parameters, space group, experimental conditions and quality measures.

Research paper thumbnail of The Neural MMO Platform for Massively Multiagent Research

ArXiv, 2021

Neural MMO is a computationally accessible research platform that combines large agent population... more Neural MMO is a computationally accessible research platform that combines large agent populations, long time horizons, open-ended tasks, and modular game systems. Existing environments feature subsets of these properties, but Neural MMO is the first to combine them all. We present Neural MMO as free and open source software with active support, ongoing development, documentation, and additional training, logging, and visualization tools to help users adapt to this new setting. Initial baselines on the platform demonstrate that agents trained in large populations explore more and learn a progression of skills. We raise other more difficult problems such as many-team cooperation as open research questions which Neural MMO is well-suited to answer. Finally, we discuss current limitations of the platform, potential mitigations, and plans for continued development.

Research paper thumbnail of Determining Question-Answer Plausibility in Crowdsourced Datasets Using Multi-Task Learning

Proceedings of the Sixth Workshop on Noisy User-generated Text (W-NUT 2020), 2020

Datasets extracted from social networks and online forums are often prone to the pitfalls of natu... more Datasets extracted from social networks and online forums are often prone to the pitfalls of natural language, namely the presence of unstructured and noisy data. In this work, we seek to enable the collection of high-quality question-answer datasets from social media by proposing a novel task for automated quality analysis and data cleaning: question-answer (QA) plausibility. Given a machine or usergenerated question and a crowd-sourced response from a social media user, we determine if the question and response are valid; if so, we identify the answer within the free-form response. We design BERT-based models to perform the QA plausibility task, and we evaluate the ability of our models to generate a clean, usable question-answer dataset. Our highestperforming approach consists of a singletask model which determines the plausibility of the question, followed by a multitask model which evaluates the plausibility of the response as well as extracts answers (Question Plausibility AUROC=0.75, Response Plausibility AUROC=0.78, Answer Extraction F1=0.665).

Research paper thumbnail of Toward Understanding the Structural Basis of Partial Agonism at the Dopamine D3 Receptor

Journal of medicinal chemistry, Jan 5, 2017

Both dopamine D3 receptor (D3R) partial agonists and antagonists have been implicated as potentia... more Both dopamine D3 receptor (D3R) partial agonists and antagonists have been implicated as potential medications for substance use disorders. In contrast to antagonists, partial agonists may cause fewer side effects since they maintain some dopaminergic tone and may be less disruptive to normal neuronal functions. Here, we report three sets of 4-phenylpiperazine stereoisomers that differ considerably in efficacy: the (R)-enantiomers are antagonists/weak partial agonists, whereas the (S)-enantiomers are much more efficacious. To investigate the structural basis of partial agonism, we performed comparative microsecond-scale molecular dynamics simulations starting from the inactive state of D3R in complex with these enantiomers. Analysis of the simulation results reveals common structural rearrangements near the ligand binding site induced by the bound (S)-enantiomers, but not by the (R)-enantiomers, that are features of partially activated receptor conformations. These receptor models b...

Research paper thumbnail of Effective Approaches to Batch Parallelization for Dynamic Neural Network Architectures

arXiv (Cornell University), Jul 8, 2017

We present a simple dynamic batching approach applicable to a large class of dynamic architecture... more We present a simple dynamic batching approach applicable to a large class of dynamic architectures that consistently yields speedups of over 10x. We provide performance bounds when the architecture is not known a priori and a stronger bound in the special case where the architecture is a predetermined balanced tree. We evaluate our approach on Johnson et al.'s recent visual question answering (VQA) result of his CLEVR dataset by Inferring and Executing Programs (IEP). We also evaluate on sparsely gated mixture of experts layers and achieve speedups of up to 1000x over the naive implementation.

Research paper thumbnail of CCDC 1489645: Experimental Crystal Structure Determination

An entry from the Cambridge Structural Database, the world's repository for small molecule cr... more An entry from the Cambridge Structural Database, the world's repository for small molecule crystal structures. The entry contains experimental data from a crystal diffraction study. The deposited dataset for this entry is freely available from the CCDC and typically includes 3D coordinates, cell parameters, space group, experimental conditions and quality measures.

Research paper thumbnail of CCDC 1489644: Experimental Crystal Structure Determination

An entry from the Cambridge Structural Database, the world's repository for small molecule cr... more An entry from the Cambridge Structural Database, the world's repository for small molecule crystal structures. The entry contains experimental data from a crystal diffraction study. The deposited dataset for this entry is freely available from the CCDC and typically includes 3D coordinates, cell parameters, space group, experimental conditions and quality measures.

Research paper thumbnail of The Neural MMO Platform for Massively Multiagent Research

ArXiv, 2021

Neural MMO is a computationally accessible research platform that combines large agent population... more Neural MMO is a computationally accessible research platform that combines large agent populations, long time horizons, open-ended tasks, and modular game systems. Existing environments feature subsets of these properties, but Neural MMO is the first to combine them all. We present Neural MMO as free and open source software with active support, ongoing development, documentation, and additional training, logging, and visualization tools to help users adapt to this new setting. Initial baselines on the platform demonstrate that agents trained in large populations explore more and learn a progression of skills. We raise other more difficult problems such as many-team cooperation as open research questions which Neural MMO is well-suited to answer. Finally, we discuss current limitations of the platform, potential mitigations, and plans for continued development.

Research paper thumbnail of Determining Question-Answer Plausibility in Crowdsourced Datasets Using Multi-Task Learning

Proceedings of the Sixth Workshop on Noisy User-generated Text (W-NUT 2020), 2020

Datasets extracted from social networks and online forums are often prone to the pitfalls of natu... more Datasets extracted from social networks and online forums are often prone to the pitfalls of natural language, namely the presence of unstructured and noisy data. In this work, we seek to enable the collection of high-quality question-answer datasets from social media by proposing a novel task for automated quality analysis and data cleaning: question-answer (QA) plausibility. Given a machine or usergenerated question and a crowd-sourced response from a social media user, we determine if the question and response are valid; if so, we identify the answer within the free-form response. We design BERT-based models to perform the QA plausibility task, and we evaluate the ability of our models to generate a clean, usable question-answer dataset. Our highestperforming approach consists of a singletask model which determines the plausibility of the question, followed by a multitask model which evaluates the plausibility of the response as well as extracts answers (Question Plausibility AUROC=0.75, Response Plausibility AUROC=0.78, Answer Extraction F1=0.665).

Research paper thumbnail of Toward Understanding the Structural Basis of Partial Agonism at the Dopamine D3 Receptor

Journal of medicinal chemistry, Jan 5, 2017

Both dopamine D3 receptor (D3R) partial agonists and antagonists have been implicated as potentia... more Both dopamine D3 receptor (D3R) partial agonists and antagonists have been implicated as potential medications for substance use disorders. In contrast to antagonists, partial agonists may cause fewer side effects since they maintain some dopaminergic tone and may be less disruptive to normal neuronal functions. Here, we report three sets of 4-phenylpiperazine stereoisomers that differ considerably in efficacy: the (R)-enantiomers are antagonists/weak partial agonists, whereas the (S)-enantiomers are much more efficacious. To investigate the structural basis of partial agonism, we performed comparative microsecond-scale molecular dynamics simulations starting from the inactive state of D3R in complex with these enantiomers. Analysis of the simulation results reveals common structural rearrangements near the ligand binding site induced by the bound (S)-enantiomers, but not by the (R)-enantiomers, that are features of partially activated receptor conformations. These receptor models b...