Search

Your search keyword '"Wang, Philip"' showing total 784 results
Start Over Author "Wang, Philip"
784 results on '"Wang, Philip"'

1. Model-Based Reward Shaping for Adversarial Inverse Reinforcement Learning in Stochastic Environments

Author

Zhan, Simon Sinong, Wu, Qingyuan, Wang, Philip, Wang, Yixuan, Jiao, Ruochen, Huang, Chao, and Zhu, Qi

Subjects
Computer Science - Machine Learning, Computer Science - Artificial Intelligence
Abstract

In this paper, we aim to tackle the limitation of the Adversarial Inverse Reinforcement Learning (AIRL) method in stochastic environments where theoretical results cannot hold and performance is degraded. To address this issue, we propose a novel method which infuses the dynamics information into the reward shaping with the theoretical guarantee for the induced optimal policy in the stochastic environments. Incorporating our novel model-enhanced rewards, we present a novel Model-Enhanced AIRL framework, which integrates transition model estimation directly into reward shaping. Furthermore, we provide a comprehensive theoretical analysis of the reward error bound and performance difference bound for our method. The experimental results in MuJoCo benchmarks show that our method can achieve superior performance in stochastic environments and competitive performance in deterministic environments, with significant improvement in sample efficiency, compared to existing baselines.

Published
2024

5. Shear response of granular packings compressed above jamming onset

Author

Wang, Philip, Zhang, Shiyun, Tuckman, Philip, Ouellette, Nicholas T., Shattuck, Mark D., and O'Hern, Corey S.

Subjects
Condensed Matter - Soft Condensed Matter
Abstract

We investigate the mechanical response of jammed packings of repulsive, frictionless spherical particles undergoing isotropic compression. Prior simulations of the soft-particle model, where the repulsive interactions scale as a power-law in the interparticle overlap with exponent $\alpha$, have found that the ensemble-averaged shear modulus $\langle G \rangle$ increases with pressure $P$ as $\sim P^{(\alpha-3/2)/(\alpha-1)}$ at large pressures. However, a deep theoretical understanding of this scaling behavior is lacking. We show that the shear modulus of jammed packings of frictionless, spherical particles has two key contributions: 1) continuous variations as a function of pressure along geometrical families, for which the interparticle contact network does not change, and 2) discontinuous jumps during compression that arise from changes in the contact network. We show that the shear modulus of the first geometrical family for jammed packings can be collapsed onto a master curve: $G^{(1)}/G_0 = (P/P_0)^{(\alpha-2)/(\alpha-1)} - P/P_0$, where $P_0 \sim N^{-2(\alpha-1)}$ is a characteristic pressure that separates the two power-law scaling regions and $G_0 \sim N^{-2(\alpha-3/2)}$. Deviations from this form can occur when there is significant non-affine particle motion near changes in the contact network. We further show that $\langle G (P)\rangle$ is not simply a sum of two power-laws, but $\langle G \rangle \sim (P/P_c)^a$, where $a \approx (\alpha -2)/(\alpha-1)$ in the $P \rightarrow 0$ limit and $\langle G \rangle \sim (P/P_c)^b$, where $b \gtrsim (\alpha -3/2)/(\alpha-1)$ above a characteristic pressure $P_c$. In addition, the magnitudes of both contributions to $\langle G\rangle$ from geometrical families and changes in the contact network remain comparable in the large-system limit for $P >P_c$., Comment: 15 pages, 19 figures

Published
2020
Full Text
View/download PDF

8. Determinants of remote measurement‐based care uptake in a safety net outpatient psychiatry department as part of learning health system transition.

Author

Aldis, Rajendra, Rosenfeld, Lisa C., Mulvaney‐Day, Norah, Lanca, Margaret, Zona, Kate, Lam, Jeffrey A., Asfour, Julia, Meltzer, Jonah C., Leff, H. Stephen, Fulwiler, Carl, Wang, Philip, and Progovac, Ana M.

Subjects
SAFETY-net health care providers, MENTAL health services, OLDER patients, ELECTRONIC health records, INSTRUCTIONAL systems, SPANISH language
Abstract

Introduction: Behavioral measurement‐based care (MBC) can improve patient outcomes and has also been advanced as a critical learning health system (LHS) tool for identifying and mitigating potential disparities in mental health treatment. However, little is known about the uptake of remote behavioral MBC in safety net settings, or possible disparities occurring in remote MBC implementation. Methods: This study uses electronic health record data to study variation in completion rates at the clinic and patient level of a remote MBC symptom measure tool during the first 6 months of implementation at three adult outpatient psychiatry clinics in a safety net health system. Provider‐reported barriers to MBC adoption were also measured using repeated surveys at one of the three sites. Results: Out of 1219 patients who were sent an MBC measure request, uptake of completing at least one measure varied by clinic: General Adult Clinic, 38% (n = 262 of 696); Substance Use Clinic, 28% (n = 73 of 265); and Transitions Clinic, 17% (n = 44 of 258). Compared with White patients, Black and Portuguese or Brazilian patients had lower uptake. Older patients also had lower uptake. Spanish language of care was associated with much lower uptake at the patient level. Significant patient‐level disparities in uptake persisted after adjusting for the clinic, mental health diagnoses, and number of measure requests sent. Providers cited time within visits and bandwidth in their workflow as the greatest consistent barriers to discussing MBC results with patients. Conclusions: There are significant disparities in MBC uptake at the patient and clinic level. From an LHS data infrastructure perspective, safety net health systems may need to address the need for possible ways to adapt MBC to better fit their populations and clinical needs, or identify targeted implementation strategies to close data gaps for the identified disparity populations. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

14. Mental Health in the Workplace : A Call to Action Proceedings From the Mental Health in the Workplace—Public Health Summit

Author

Goetzel, Ron Z., Roemer, Enid Chung, Holingue, Calliope, Fallin, Daniele, McCleary, Katherine, Eaton, William, Agnew, Jacqueline, Azocar, Francisca, Ballard, David, Bartlett, John, Braga, Michael, Conway, Heidi, Crighton, K. Andrew, Frank, Richard, Jinnett, Kim, Keller-Greene, Debra, Rauch, Sara Martin, Safeer, Richard, Saporito, Dick, Schill, Anita, Shern, David, Strecher, Victor, Wald, Peter, Wang, Philip, and Mattingly, C. Richard

Published
2018

17. The Translational Science Training Program at NIH: Introducing Early Career Researchers to the Science and Operation of Translation of Basic Research to Medical Interventions

Author

Gilliland, C. Taylor, Sittampalam, G. Sitta, Wang, Philip Y., and Ryan, Philip E.

Abstract

Translational science is an emerging field that holds great promise to accelerate the development of novel medical interventions. As the field grows, so does the demand for highly trained biomedical scientists to fill the positions that are being created. Many graduate and postdoctorate training programs do not provide their trainees with sufficient education to take advantage of this growing employment sector. To help better prepare the trainees at the National Institutes of Health for possible careers in translation, we have created the Translational Science Training Program (TSTP). The TSTP is an intensive 2- to 3-day training program that introduces NIH postdoctoral trainees and graduate students to the science and operation of turning basic research discoveries into a medical therapeutic, device or diagnostic, and also exposes them to the variety of career options in translational science. Through a combination of classroom teaching from practicing experts in the various disciplines of translation and small group interactions with pre-clinical development teams, participants in the TSTP gain knowledge that will aid them in obtaining a career in translational science and building a network to make the transition to the field.

Published
2017
Full Text
View/download PDF

49. Anomaly Detection using a Deep Learning Multi-layer Perceptron to Mitigate the Risk of Rogue Trading

Author

Hedström, Erik and Wang, Philip

Subjects
Machine Learning, Computer and Information Sciences, Deep Learning, Rogue Trading, Multi-layer Perceptron (MLP), Data- och informationsvetenskap, Neural network
Abstract

The term Rogue Trading is defined as the activity of someone at a financial organisation losing a large amount of money in bad or illegal transactions and trying to hide this. The activity of Rogue traders exposes financial organisations to huge risks and may lead to the organisation collapsing, which will affect other stakeholders like, for example, the customers. In order to detect potential Rogue Trading cases, Control Systems that monitor the employees and the positions they take on financial markets must exist. In this study, a two-step control system is suggested to monitor the margins on Foreign exchange (FX) Forwards traded by employees at the Swedish bank Skandinaviska Enskilda Banken (SEB). The first step in the control system uses a Deep Learning neural network trained on transactional data to predict the margin. The errors of the predictions versus the actual values are then in the second step of the control system used to find outliers which should be flagged for further investigation due to a too high deviation. The results show that the model hopefully can decrease the number of false positives yielded by the current Control Systems at SEB and thus reduce manual inspection of flagged transactions. Termen Rouge Trading definieras som en aktivitet där någon på en finansiell institution förlorar stora mängder pengar i dåliga eller illegala transaktioner och försöker dölja detta. Detta är något som skapar enorma risker för finansiella institutioner och som kan förorsaka organisationens kollaps, som kan påverka intressenter som till exempel kunder. För att upptäcka potentiella företeelser av Rouge Trading så måste kontrollsystem som övervakar anställda och deras positioner existera. I denna studie föreslås och presenteras ett tvåstegs-system för att övervaka marginaler vid terminsaffärer i utländsk valuta vid Skandinaviska Enskilda Banken (SEB). Det första steget i kontrollsystemet använder ett neuralt närverk tränat på data från transaktioner för att prediktera en marginal. Differenserna mellan prediktionen och det faktiska värdet används för att finna outliers vilka borde flaggas för vidare undersökning. Resultaten visar att modellen förhoppningsvis kan minska antalet falska positiva som det nuvarande kontrollsystemet ger på SEB, något som således kan minska den manuella inspektionen av flaggade transaktioner.

Published
2021

Catalog

Books, media, physical & digital resources
See catalog results