Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Van Roy
Toward Robust InP Laser Lift-Off: Energy-Dependent Absorption Shift and Self-Absorption under Single-Pulse 1064 nm Irradiation
Heterogeneous integration of InP on silicon is essential for the next generation of electronic systems by enabling high-performance photonics and radio frequency devices. Near-infrared Laser Lift-Off (LLO) is a promising route to enable the cost-effective transfer of InP-based device layers onto Si platforms. However, its underlying mechanisms remain poorly understood. To clarify these laser-material interactions, we investigate an InP/InGaAs heterostructure under single-pulse 1064 nm irradiation.
Computer Science > Artificial Intelligence
In July 2026, OpenAI's agents coordinated over channels outside their intended environment to breach Hugging Face's secured infrastructure. Could existing alignment testing practices have foreseen this incident? If not, what needs to change? We explore these questions. First, we identify the misaligned behaviors that caused this incident.
Computer Science > Machine Learning
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Computer Science > Artificial Intelligence
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2507.21479] Capacity-Constrained Continual Learning
arXiv:2507.21479 (cs) View PDF HTML (experimental) Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Theory (cs.IT); Systems and Control (eess.SY); Machine Learning (stat.ML) Cite as: arXiv:2507.21479 [cs.LG] (or arXiv:2507.21479v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2507.21479 Submission history From: Zheng Wen [ view email] [v1] Tue, 29 Jul 2025 03:47:22 UTC (600 KB) Bibliographic Tools Bibliographic Explorer Toggle Bibliographic Explorer...
[2507.12041] Granular feedback merits sophisticated aggregation
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Interferon-gamma receptor signaling regulates innate immunity during Staphylococcus aureus craniotomy infection - Journal of Neuroinflammation
Our previous work demonstrated that CD4+ Th1 and Th17 cells are critical for bacterial containment during craniotomy infection [27]; however, both populations produce TNF as well as proinflammatory mediators unique to Th1 (IFN-γ and IL-2) and Th17 cells (IL-17, IL-21, and IL-22) [56,57,58] that could impact infectious outcomes.
[2410.22690] Choice between Partial Trajectories
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
The Potential Use of Targeted Proteomics and Metabolomics for the Identification and Monitoring of Diabetic Kidney Disease
All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess.
The Need for a Big World Simulator: A Scientific Challenge for Continual Learning
Want to see access to research regardless of disability? Sign up for the arXiv Accessibility Forum in September and Learn more.
Tumor necrosis factor regulates leukocyte recruitment but not bacterial persistence during Staphylococcus aureus craniotomy infection - Journal of Neuroinflammation
Craniotomy infection was established in 8–10 week-old male and female C57BL/6J wild type (WT; RRID: IMSR_JAX:002365), Tnf−/− (TNF KO; RRID: IMSR_JAX:00300), Tnfr1−/− (TNFR1 KO; RRID: IMSR_JAX:003242), Tnfr2−/− (TNFR2 KO; RRID: IMSR_JAX:002620), or Pad4−/− (PAD4 KO; RRID: IMSR_JAX:030315) mice as previously described [2, 7, 10, 11]. Briefly, a surgical plane of anesthesia was achieved by i.p. ketamine and xylazine administration before exposing the cranium with a midline scalp incision.
[2407.12185] Satisficing Exploration for Deep Reinforcement Learning
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2402.00396] Efficient Exploration for LLMs
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2401.13239] Adaptive Crowdsourcing Via Self-Supervised Learning
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2310.07786] Non-Stationary Contextual Bandit Learning via Neural Predictive Ensemble Sampling
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2308.11958] Maintaining Plasticity in Continual Learning via Regenerative Regularization
[Submitted on 23 Aug 2023 ( v1 ), last revised 24 Oct 2024 (this version, v3)] Title:Maintaining Plasticity in Continual Learning via Regenerative Regularization View a PDF of the paper titled Maintaining Plasticity in Continual Learning via Regenerative Regularization, by Saurabh Kumar and 2 other authors View PDF HTML (experimental) Abstract:In continual learning, plasticity refers to the ability of an agent to quickly adapt to new information.
Mission and role modelling in producing a fit‐for‐purpose rural health workforce: perspectives from an international community of practice
Med J Aust 2023; 219 (3): S20-S26. || doi: 10.5694/mja2.52020 Published online: 7 August 2023 Inequities in the distribution of human resources in health around the world have long been a topic of concern and discussion.
Dilemmas in the Management of an Infant with Neuroblastoma Metastasized to the Muscles
The definition of Stage Ms does not distinguish between the lateralization and pattern of lymph node metastases. Yet the presence of MYCN-amplification in the presence of a stage Ms staging determines an upstaging in risk stratification [14]. This definition is supported by findings reported in the study by Morgenstern et al. [17]. In our case, only the age at diagnosis was favorable.
[2307.11046] A Definition of Continual Reinforcement Learning
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2307.04345] Continual Learning as Computationally Constrained Reinforcement Learning
arXiv:2307.04345 (cs) [Submitted on 10 Jul 2023 (v1), last revised 26 Jun 2025 (this version, v3)] View PDF HTML (experimental) Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2307.04345 [cs.LG] (or arXiv:2307.04345v3 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2307.04345 Submission history From: Saurabh Kumar [ view email] [v1] Mon, 10 Jul 2023 05:06:41 UTC (2,188 KB) [v2] Sun, 20 Aug 2023 19:58:49 UTC (2,411 KB) [v3] Thu, 26 Jun 2025 16:08:44...
[2306.14834] Scalable Neural Contextual Bandit for Recommender Systems
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2109.12509] Deep Exploration for Recommendation Systems
[Submitted on 26 Sep 2021 ( v1 ), last revised 30 Jul 2023 (this version, v4)] Title:Deep Exploration for Recommendation Systems Download a PDF of the paper titled Deep Exploration for Recommendation Systems, by Zheqing Zhu and 1 other authors Download PDF Abstract: Modern recommendation systems ought to benefit by probing for and learning from delayed feedback. Research has tended to focus on learning from a user's response to a single recommendation.
[2305.11455] Shattering the Agent-Environment Interface for Fine-Tuning Inclusive Language Models
arXiv:2305.11455 (cs) Download a PDF of the paper titled Shattering the Agent-Environment Interface for Fine-Tuning Inclusive Language Models, by Wanqiao Xu and 3 other authors Download PDF Submission history From: Wanqiao Xu [ view email] [v1] Fri, 19 May 2023 06:21:15 UTC (1,425 KB) Bibliographic Tools Bibliographic Explorer Toggle Bibliographic Explorer () Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media DagsHub...
[2107.08924] Epistemic Neural Networks
[Submitted on 19 Jul 2021 ( v1 ), last revised 17 May 2023 (this version, v8)] Title:Epistemic Neural Networks Download a PDF of the paper titled Epistemic Neural Networks, by Ian Osband and 6 other authors Download PDF Abstract: Intelligence relies on an agent's knowledge of what it does not know. This capability can be assessed based on the quality of joint predictions of labels across multiple inputs.
[2302.03319] Leveraging Demonstrations to Improve Online Learning: Quality Matters
[Submitted on 7 Feb 2023 ( v1 ), last revised 16 May 2023 (this version, v3)] Title:Leveraging Demonstrations to Improve Online Learning: Quality Matters Download a PDF of the paper titled Leveraging Demonstrations to Improve Online Learning: Quality Matters, by Botao Hao and 4 other authors Download PDF Abstract: We investigate the extent to which offline demonstration data can improve online learning. It is natural to expect some improvement, but the question is how, and by how much?
IL-10 production by granulocytes promotes Staphylococcus aureus craniotomy infection - Journal of Neuroinflammation
References Fernandez-de Thomas RJ, De Jesus O. Craniotomy. StatPearls. Treasure Island (FL)2021. Jimenez-Martinez E, Cuervo G, Hornero A, Ciercoles P, Gabarros A, Cabellos C, et al. Risk factors for surgical site infection after craniotomy: a prospective cohort study. Antimicrob Resist Infect Control. 2019;8:69. PubMed PubMed Central Google Scholar McClelland S 3rd, Hall WA. Postoperative central nervous system infection: incidence and associated factors in 2111 neurosurgical procedures. Clin Infect Dis.
[2211.01568] Fine-Tuning Language Models via Epistemic Neural Networks
[Submitted on 3 Nov 2022 ( v1 ), last revised 10 May 2023 (this version, v2)] Title:Fine-Tuning Language Models via Epistemic Neural Networks Download a PDF of the paper titled Fine-Tuning Language Models via Epistemic Neural Networks, by Ian Osband and 5 other authors Download PDF Abstract: Language models often pre-train on large unsupervised text corpora, then fine-tune on additional task-specific data. However, typical fine-tuning schemes do not prioritize the examples that they tune on.
[2305.03263] Bayesian Reinforcement Learning with Limited Cognitive Load
All biological and artificial agents must learn and make decisions given limits on their ability to process information. As such, a general theory of adaptive behavior should be able to account for the complex interactions between an agent's learning history, decisions, and capacity constraints.
[2103.04047] Reinforcement Learning, Bit by Bit
[Submitted on 6 Mar 2021 ( v1 ), last revised 4 May 2023 (this version, v8)] Title:Reinforcement Learning, Bit by Bit Download a PDF of the paper titled Reinforcement Learning, Bit by Bit, by Xiuyuan Lu and 5 other authors Download PDF Abstract: Reinforcement learning agents have demonstrated remarkable achievements in simulated environments. Data efficiency poses an impediment to carrying this success over to real environments.
[1705.07347] Ensemble Sampling
[Submitted on 20 May 2017 ( v1 ), last revised 25 Apr 2023 (this version, v4)] Title:Ensemble Sampling Download a PDF of the paper titled Ensemble Sampling, by Xiuyuan Lu and 1 other authors Download PDF Abstract: Thompson sampling has emerged as an effective heuristic for a broad range of online decision problems. In its basic form, the algorithm requires computing and sampling from a posterior distribution over models, which is tractable only for simple special cases.
[2302.12202] A Definition of Non-Stationary Bandits
[Submitted on 23 Feb 2023 ( v1 ), last revised 28 Jul 2023 (this version, v2)] Title:A Definition of Non-Stationary Bandits Download a PDF of the paper titled A Definition of Non-Stationary Bandits, by Yueyang Liu and 2 other authors Download PDF Abstract: Despite the subject of non-stationary bandit learning having attracted much recent attention, we have yet to identify a formal definition of non-stationarity that can consistently distinguish non-stationary bandits from stationary ones.
[2212.01365] An Information-Theoretic Analysis of Compute-Optimal Neural Scaling Laws
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
[2205.01970] Non-Stationary Bandit Learning via Predictive Sampling
[Submitted on 4 May 2022 ( v1 ), last revised 17 Jul 2023 (this version, v6)] Title:Non-Stationary Bandit Learning via Predictive Sampling Download a PDF of the paper titled Non-Stationary Bandit Learning via Predictive Sampling, by Yueyang Liu and 2 other authors Download PDF Abstract: Thompson sampling has proven effective across a wide range of stationary bandit environments. However, as we demonstrate in this paper, it can perform poorly when applied to non-stationary environments.
Evaluating Predictive Distributions: Does Bayesian Deep Learning Work?
Global Survey In just 3 minutes, help us better understand how you perceive arXiv. Take the survey
[1908.03568v1] Behaviour Suite for Reinforcement Learning
Authors:Ian Osband, Yotam Doron, Matteo Hessel, John Aslanides, Eren Sezener, Andre Saraiva, Katrina McKinney, Tor Lattimore, Csaba Szepezvari, Satinder Singh, Benjamin Van Roy, Richard Sutton, David Silver, Hado Van Hasselt (Submitted on 9 Aug 2019) Abstract: This paper introduces the Behaviour Suite for Reinforcement Learning, or bsuite for short.
[1707.02038] A Tutorial on Thompson Sampling
arXiv:1707.02038 (cs) [Submitted on 7 Jul 2017 (v1), last revised 14 Jul 2020 (this version, v3)] View PDF Subjects: Machine Learning (cs.LG) Cite as: arXiv:1707.02038 [cs.LG] (or arXiv:1707.02038v3 [cs.LG] for this version) https://doi.org/10.48550/arXiv.1707.02038 Journal reference: Foundations and Trends in Machine Learning, Vol. 11, No. 1, pp.
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Van
Contact Van, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack