Pith. sign in

REVIEW 4 major objections 6 minor 100 references

Fast and Robust: Task Sampling with Posterior and Diversity Synergies for Adaptive Decision-Makers in Randomized Environments

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Posterior sampling plus a diversity penalty makes active task sampling stable at 64x candidate pools and improves CVaR adaptation robustness across Meta-RL and domain-randomization benchmarks.

desk verdict Empirically promising task-sampling method whose theory section is mostly cosmetic; the robustness claim should survive review but needs surrogate validation and proof fixes. read the letter →

arxiv 2504.19139 v3 pith:656Z7ZCE submitted 2025-04-27 cs.LG cs.AIstat.ML

Yun Qu , Qi Cheems Wang , Yixiu Mao , Yiqin Lv , Xiangyang Ji This is my paper · ORCID
classification cs.LGcs.AIstat.ML
keywords tasksamplingdomainrandomizationmetareinforcementlearningconditionalvalue-at-riskposteriordiversityregularizationinfinitelymany-armedbanditrisk-aversedecision-making
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper confronts a cost problem in risk-averse reinforcement learning: to make an adaptive policy robust to the worst environments, training must emphasize difficult tasks, but identifying them normally requires expensive policy evaluation. It studies a cheaper route, robust active task sampling, in which a risk-predictive model scores many candidate tasks and a sampler picks the batch to train on. The paper's central claim is that two modifications—drawing the acquisition scores from a single posterior sample instead of a UCB estimate, and penalizing selected subsets that lack task diversity—make this route stable and effective, allowing a 64x larger candidate pool without performance collapse. If correct, robust adaptation becomes considerably cheaper for sim-to-real robotics, Meta-RL, and other randomized-environment tasks, with worst-tail returns comparable to or better than risk-averse baselines. The paper also offers a theoretical lens: the sampling loop is modeled as a Markov decision process and an infinite-armed bandit, giving a unified view of existing methods and the new one.

What carries the argument

The load-bearing object is the task-selection MDP collapsed into an infinitely many-armed bandit (i-MAB): each arm is a size-B subset of the candidate pool, the state is the current policy θ, and the step reward is the CVaR decrease from optimizing on that subset. On top of this, the method replaces MPTS's UCB acquisition with a single posterior draw—sample latent z from the approximate posterior, draw predicted risk for every candidate task, then pick the B-subset maximizing sum of sampled risk plus γ times pairwise diversity of task identifiers (Eq. 12). The concentration proposition (3.3) is the mechanism that explains why pure top-B selection fails as the candidate pool grows: for a unimodal risk function, the top-B set converges to the maximizing neighborhood. Diversity regularization counteracts exactly that.

What would settle it

A decisive test: on a Meta-RL benchmark, corrupt the risk-predictive model by permuting its predicted risks across candidate tasks, keeping the sampler otherwise intact; if PDTS's CVaR advantage over ERM survives the permutation, the diversity term alone is doing the work, whereas if it collapses, the surrogate ranking is load-bearing. A second, cheaper check is to measure the selected-task variance as the candidate pool grows: PDTS predicts that top-B selection concentrates while diversity-regularized selection does not.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the standard way of picking robust tasks—rank all candidates by predicted risk and keep the top-B—has a structural flaw: as the candidate pool grows, those top-B tasks cluster in a tiny neighborhood of the hardest point, over-optimizing one region and degrading the policy. The paper proves a concentration bound for this phenomenon (Proposition 3.3) and then shows that adding a diversity penalty on the selected subset while replacing UCB scores with one sample from the posterior (Eq. 12) removes the collapse. The resulting PDTS achieves what the paper calls nearly worst-case optimization, and in experiments it improves CVaR robustness over MPTS, ERM, GDRM, and DRM on Meta-RL and physical and visual domain-randomization benchmarks, while keeping computational cost comparable to ERM.

Load-bearing premise

The whole scheme depends on the risk-predictive model from MPTS ranking arbitrary candidate tasks by difficulty accurately enough, and on that ranking staying trustworthy when the sampler's own selections shape the next round of training data; if the surrogate scores are biased, both the posterior draw and the diversity filter are selecting on noise.

Editorial extensions

If this is right

  • At a 64x candidate pool, PDTS avoids the concentration collapse that MPTS exhibits at 8x, letting the sampler screen far more tasks without extra environment interaction.
  • Across the benchmarked Meta-RL and DR tasks, PDTS's CVaR0.9 returns exceed MPTS, ERM, GDRM, and DRM, with the margin growing as the risk level α approaches 1.
  • On LunarLander, PDTS shows the smallest performance drop when moving to out-of-distribution identifier ranges, especially on the hardest shifted tasks.
  • In Pusher and LunarLander, PDTS reaches ERM-level average returns with roughly 2.4x and 1.3x fewer training steps, respectively.
  • PDTS works with both MAML and PEARL meta-learners and with TD3 and PPO backbones in domain randomization, so the gain is not tied to one optimizer.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the diversity screen can only be as good as the posterior draw, improving the risk-predictive model—flagged by the paper as future work—should compound PDTS's gains rather than merely add to them.
  • The concentration diagnosis applies to any unimodal scoring function, so the same collapse-and-remedy pattern should appear in other batch-selection settings where a noisy surrogate is followed by top-B acquisition; this is a testable transfer, not something the paper demonstrates.
  • A direct prediction: PDTS's advantage should shrink when task identifiers become high-dimensional or the risk function becomes rough, because both the surrogate ranking and pairwise-distance diversity lose signal in that regime.
  • The i-MAB framing leaves the acquisition rule pluggable, so alternative optimistic strategies could be substituted into the same regularized subset-selection loop; the paper does not explore this.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This paper proposes Posterior and Diversity Synergized Task Sampling (PDTS), an active task-sampling method for risk-averse Meta-RL and domain randomization. The authors frame robust active task sampling as a task-selection MDP and an infinitely many-armed bandit (i-MAB), argue that MPTS is a UCB-guided special case, diagnose a concentration issue when the candidate pool is enlarged, and propose replacing UCB with posterior sampling plus a diversity regularizer. The empirical section reports improved CVaR returns over ERM, MPTS, GDRM, and DRM on Meta-RL benchmarks, physical robotics domain randomization, visual ManiSkill3 scenarios, and sinusoid regression, with seven seeds and standard errors.

Significance. If the empirical results hold, the contribution is practically valuable: PDTS is easy to implement (one posterior sample per candidate plus a standard diversity maximization step), inherits MPTS's amortized evaluation, and appears to avoid the performance collapse MPTS suffers at larger candidate-pool sizes. The evaluation is unusually broad for this line of work, including Meta-RL with MAML and PEARL backbones, three domain-randomization suites, OOD tests, ablations, and reproducible code/project artifacts. The main weakness is that the theoretical claims, especially the 'nearly worst-case' guarantee, are not actually derived, and the robustness of the whole pipeline rests on an inherited risk-prediction model whose ranking quality on unseen hard tasks is not validated under the self-selection loop. These issues are fixable by re-scoping the theory and adding validation experiments, so the work has solid potential after major revision.

major comments (4)
  1. [Sec. 3.2, Eq. (11), Appendix B.5] Proposition 3.4 attaches the label 'nearly worst-case optimization' to the regularized objective, but the proof in Appendix B.5 only says that diversity regularization 'perturbs the worst arm selection.' With gamma > 0, the selected subset is by construction not the worst subset, and no quantitative bound is given on the gap between the value of Eq. (11) and CVaR_{1-B/hat(B)} or the worst-case subset value. As stated, the proposition is a naming choice rather than a derived guarantee. Either remove the proposition, or prove an explicit approximation bound (e.g., via submodularity of S) and state the required smoothness assumptions.
  2. [Appendix B.4 (Proposition 3.3)] The proof contains an incorrect binomial expression: the probability that at least B of the hat(B) i.i.d. samples fall in the epsilon-neighborhood is 1 - sum_{j=0}^{B-1} C(hat(B), j) p_epsilon^j (1-p_epsilon)^{hat(B)-j}, which is not the displayed sum. The monotonicity argument also compares C(hat(B), i-1) with C(hat(B)+1, i-1) and requires p_epsilon < (hat(B)-i+2)/(hat(B)+1) for each i, not a single condition independent of i; there is additionally a typo 'n+1' in the final condition. Since Proposition 3.3 is the theoretical diagnosis of the concentration issue that motivates the diversity regularizer, this part of the theory needs to be corrected or replaced with a cleaner argument.
  3. [Sec. 2.2, Eq. (5); Algorithm 1; Fig. 6(c); Limitations] The central empirical claim that PDTS improves CVaR robustness by actively selecting hard tasks depends on the inherited risk model p(ell | tau, H; theta) ranking arbitrary candidate tasks by difficulty. In Algorithm 1 only the B selected tasks receive exact labels; the remaining hat(B)-B candidates are scored by extrapolation, and those same scores feed the next selection, so the model is trained on its own self-selected history and the unvisited hard-task region is never validated. The paper's evidence is in-distribution Pearson correlation (Fig. 6(c)), which does not test ranking quality on fresh random candidates, and the Limitations section concedes that the required smoothness/scoring assumptions 'may not always hold.' Without an out-of-distribution or cold-start validation of the surrogate ranking, the reported advantage over ERM remains conditional on an untested generalization property. I recommend adding such a validation (e.g., periodically score a fresh random task set, evaluate the top/Bottom predicted tasks exactly, and report rank correlation or top-k precision).
  4. [Sec. 3.1, Eq. (9); Appendix B.3, Lemma B.1] The Bellman optimality statement in Eq. (9) is not generally valid as written: the optimal suffix from time t depends on the state theta_t reached by the first t decisions, so the right-hand side should maximize over the remaining policy conditional on theta_t rather than reusing the same fixed suffix. Lemma B.1 also asserts that Top-B risk selection maximizes the CVaR decrease using a Taylor expansion that assumes equal gradient norms for the unbiased and biased objectives and ignores higher-order terms; these are non-trivial assumptions not stated in the proposition. Because the i-MAB formulation is advertised as a main theoretical contribution, these steps should be either proved rigorously or presented explicitly as heuristics.
minor comments (6)
  1. [Appendix B.4] There are typos 'step-wsie' and 'n+1' in the proof of Proposition 3.3; please correct them and standardize the notation for hat(B), B, and p_epsilon.
  2. [Fig. 3 and Sec. 4.2] The x-axis label in Fig. 3 reads 'T ask Identifier' and should read 'Task Identifier'; in Sec. 4.2, 'Lundarlander' should be 'LunarLander'.
  3. [References] References Wang et al. 2025a and Wang et al. 2025b share the same arXiv identifier 2501.11039; please clarify whether these are the same manuscript and cite distinct works accordingly.
  4. [Algorithms 3 and 5] Algorithms 3 and 5 duplicate Algorithm 1 verbatim; this repetition is unnecessary and could be replaced by a pointer to Algorithm 1.
  5. [Fig. 6(d)] Fig. 6(d) reports relative memory and clock time; please state the reference point in the caption and specify whether these are per-iteration or total costs.
  6. [Sec. 4.2] The claim that PDTS 'outperforms ERM by more than 8% on all benchmarks in CVaR0.9, and by as much as 73% on LunarLander' should state the exact CVaR level and include the corresponding standard errors for those reported percentages.

Circularity Check

2 steps flagged · score 5.0 of 10

The 'nearly worst-case' proposition is a self-defined label for diversity-perturbed selection, and the surrogate-ranking premise is inherited from the authors' own MPTS paper; the empirical comparisons remain independent.

  1. self definitional [Section 3.2, Proposition 3.4 and its proof; Eq. (11)]
    "Proposition 3.4 (Nearly Worst-Case Optimization with PDTS). When ˆB grows large enough, optimizing the subset from Eq. (11) achieves nearly worst-case optimization. ... Since the involvement of the diversity regularization perturbs the worst arm selection, this induces the nearly worst-case optimization in PDTS."

    The predicate being proved ('nearly worst-case optimization') is defined by the same perturbation that Eq. (11) introduces. For any γ > 0, the argmax of A(TB) + γS[{d(τi,τj)}] is, by construction, not the argmax of A(TB) unless S is constant over feasible subsets. Therefore the statement that PDTS achieves 'nearly worst-case' optimization is a restatement of the objective's diversity term, not a derived bound or theorem. The proof's only step is: diversity perturbs worst-arm selection, therefore it is nearly worst-case. No quantitative notion of 'nearly' is provided, so the proposition names a design choice rather than deriving a robustness guarantee.

  2. self citation load bearing [Section 2.2, RATS Preliminaries; Contributions, Section 1; Appendix A.5; Limitations, Section 5]
    "In particular, the empirical evidence in (Wang et al., 2025b) Fig. 5 validates its risk predictive model’s feasibility of approximately scoring MDPs’ difficulties with high Pearson correlation coefficients between the model predictive ones and exact evaluation."

    The paper's robustness advantage over ERM depends on the inherited risk model scoring the difficulty of candidate tasks that were never exactly evaluated. The external warrant cited for this scoring capability is the authors' own prior MPTS paper, and Appendix A.5 states 'we adopt the same risk predictive model design proposed in MPTS.' The in-paper Fig. 6(c) PCC evidence is on validation tasks, not on fresh candidates under the self-selection loop. The paper itself concedes in Limitations that the model only 'roughly' scores task difficulties and that the assumptions 'may not always hold.' Thus a load-bearing premise is justified by a same-group citation plus in-distribution correlation, with no independent test of top-of-difficulty ranking quality on unvisited tasks.

full rationale

The empirical core of PDTS—CVaR returns on Meta-RL, physical DR, and visual benchmarks against MPTS/ERM/GDRM/DRM—is not circular: these are external benchmark results, and the ablations independently show the concentration collapse and the contribution of the diversity and posterior components. However, two load-bearing pieces are weaker than they appear. Proposition 3.4's 'nearly worst-case' claim is self-definitional: it names diversity-perturbed selection as 'nearly worst-case' without deriving any approximation guarantee. And the difficulty surrogate on which the whole RATS pipeline relies is inherited from the authors' own MPTS paper, with only in-distribution PCC validation in this paper; the Limitations section explicitly acknowledges this reliance and the possibility that the assumptions fail. These issues make the robustness conclusion conditional, but they do not make the benchmark comparisons themselves circular, so the score is moderate.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The task-selection MDP M and the i-MAB are analytical abstractions used to frame the method; no new physical or ontological entities are postulated, so there are no invented entities with or without independent evidence.

free parameters (4)
  • diversity regularization weight gamma = 1 (ablated over {0,1,2,4,8})
    Appears in Eq. (11); ablation (Fig. 13a) shows sensitivity, with 'increasing gamma appropriately' for low-dimensional identifiers, i.e., per-benchmark tuning.
  • candidate batch multiple hat B / B = 64x for PDTS (vs 1.5x to 25x for MPTS)
    Chosen by hand; the method's benefits depend heavily on this scale (Fig. 13d), and it is part of the comparison protocol.
  • KL penalty weight beta in GELBO = not stated numerically
    Appears in Eq. (5), inherited from MPTS's beta-VAE objective; a hyperparameter of the risk predictive model.
  • risk model update steps K and learning rate lambda2 = 15000 steps; 5e-4 or 5e-3 per scenario
    Standard training hyperparameters for the surrogate model, not constants fitted to the target robustness claim.
assumptions (6)
  • domain assumption Assumption 1: adaptation risk is beta_tau-Lipschitz in theta and beta_theta-Lipschitz in tau (Eq. 17).
    Used in Lemma B.1's Taylor expansion to justify the top-B subset maximizing the step reward; not verified empirically in the paper.
  • domain assumption Assumption 2: adaptation risk is bounded by ell_max (Eq. 18).
    Stated as a common assumption needed for both MPTS and PDTS.
  • ad hoc to paper Assumption 3: predicted adaptation risk values follow an implicit Gaussian distribution (Appendix B.3).
    Introduced to derive the approximate-UCB form of MPTS; the Gaussianity of Monte-Carlo-estimated risk values is not tested.
  • domain assumption The function f(tau) in Proposition 3.3 is unimodal and continuous on R^d.
    Needed for the concentration claim; real adaptation risk functions need not be unimodal.
  • domain assumption Candidate tasks are sampled i.i.d. from p(tau), and the task-selection MDP's transition is a deterministic smooth optimizer F (Appendix B.2).
    Underlies the i-MAB reduction; real policy optimization is stochastic and non-smooth.
  • domain assumption The risk predictive model p(ell|tau,H) scores task difficulties with sufficient accuracy to rank the candidate pool.
    Load-bearing for the whole PDTS pipeline (Section 2.2, Algorithm 1); acknowledged in Limitations, validated only in-distribution (Fig. 6c).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fast and Robust: Task Sampling with Posterior and Diversity Synergies for Adaptive Decision-Makers in Randomized Environments." pith.science (2026). https://pith.science/paper/656Z7ZCE

@misc{pith2026250419139,
  author       = {Pith},
  title        = {Pith review of: Fast and Robust: Task Sampling with Posterior and Diversity Synergies for Adaptive Decision-Makers in Randomized Environments},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/656Z7ZCE}},
  note         = {Machine review of arXiv:2504.19139}
}
read the original abstract

Task robust adaptation is a long-standing pursuit in sequential decision-making. Some risk-averse strategies, e.g., the conditional value-at-risk principle, are incorporated in domain randomization or meta reinforcement learning to prioritize difficult tasks in optimization, which demand costly intensive evaluations. The efficiency issue prompts the development of robust active task sampling to train adaptive policies, where risk-predictive models are used to surrogate policy evaluation. This work characterizes the optimization pipeline of robust active task sampling as a Markov decision process, posits theoretical and practical insights, and constitutes robustness concepts in risk-averse scenarios. Importantly, we propose an easy-to-implement method, referred to as Posterior and Diversity Synergized Task Sampling (PDTS), to accommodate fast and robust sequential decision-making. Extensive experiments show that PDTS unlocks the potential of robust active task sampling, significantly improves the zero-shot and few-shot adaptation robustness in challenging tasks, and even accelerates the learning process under certain scenarios. Our project website is at https://thu-rllab.github.io/PDTS_project_page.

Figures

Figures reproduced from arXiv: 2504.19139 by the authors.

Figure 1
Figure 1. (a) General RATS in risk-averse decision-making. The pipeline involves amortized evaluation of task difficulties, robust subset selection, policy optimization in the MDP batch, and risk predictive models’ update. [fire: updates; snow: evaluation] (b) PDTS as a RATS method. PDTS treats task subsets as bandit arms, evaluates values through posterior sampling, and solves a regularized problem. principles in MPTS. The a… view at source ↗
Figure 2
Figure 2. Task Robust Episodic Learning as a Task-Selection MDP. Task Robust Episodic Learning as a MDP. The primary insight lies in a finite-horizon Markov decision process (Put￾erman, 2014), denoted by M =< S, A, P, R >. Here, we specify the essential components of M as: • State Space. M treats the feasible machine learner’s pa￾rameter, such as Meta-RL policies, as the reachable state, i.e., S = {θ ∈ Θ}; • Action Space. The… view at source ↗
Figure 3
Figure 3. MPTS’s Performance Collapse with Greater Bˆ. We report the performance collapses of MPTS on Walker2dVel in the case Bˆ = 8B. The task sampling frequency reveals the presence of the concentration issue. cost is significantly lower than that of agent-environment interactions and policy optimization in Meta-RL or DR. Unfortunately, MPTS might encounter performance col￾lapse with greater Bˆ , as indicated in [PITH_FULL… view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Meta-RL Results. The top depicts the cumulative return curves for CVaR0.9 validation MDPs during meta-training; the middle shows the average cumulative returns curves during meta-training; and the bottom presents the meta-testing results with various α. baselines as de…
Figure 5
Figure 5. Figure 5: Physical Robotics DR Results. (a) The top shows the cumulative return curves for CVaR0.9 validation MDPs during training; the middle displays the average cumulative return curves across all validation MDPs during training; and the bottom presents the test results at va…
Figure 6
Figure 6. Figure 6: Visual Robotics DR results. (a) Illustrations of two visual DR scenarios. (b) Curves of the average success ratio and the CVaR0.5 success ratio on validation tasks during training. (c) Training curves of PCC values between predicted and true episode returns. (d) Memory…
Figure 7
Figure 7. Figure 7: The probabilistic graphical model of the risk predictive model in (Wang et al., 2025b), where gray units denote observed variables with the white as unobservable ones. The solid directed lines depict the generative model, and the dashed directed lines indicate the reco…
Figure 8
Figure 8. Figure 8: Illustration of the neural architecture of the risk predictive model in (Wang et al., 2025b). The risk predictive model follows an encoder-decoder structure which encodes the batch [τt,i, ℓt,i] into a latent variable z and then decodes it into predicted adaptation risk…
Figure 9
Figure 9. Figure 9: Illustrations of three types of robots in Meta-RL based on the Mujoco. (a) Reacher, (b) HalfCheetah, and (c) Walker2d [PITH_FULL_IMAGE:figures/full_fig_p024_9.png]
Figure 10
Figure 10. Figure 10: Illustrations of three physical robotics domain randomization scenarios: (a) LunarLander, (b) Pusher, and (c) ErgoReacher. The illustration of ErgoReacher is adapted from (Mehta et al., 2020). supplemented by a control cost to encourage robustness. As the task distrib…
Figure 11
Figure 11. Figure 11: Illustrations of two visual robotics domain randomization scenarios: controlling (a) a table-top robotic arm and (b) a quadruped robot, operating under randomized lighting conditions and varying goal locations, respectively. • ErgoReacher involves a 4-DoF robotic arm …
Figure 12
Figure 12. Figure 12: Few-Shot Sinusoid Regression Results. (a) Illustration of the sinusoid regression problem, where the task identifier τ consists of the amplitude and phase [a, b]. (b) Curves of averaged MSEs on the validation task set for all methods during meta-training. (c) MSE valu…
Figure 13
Figure 13. Figure 13: (a) Meta-testing results trained with different hyperparameter, γ. (b) Ablation studies on key components, including posterior sampling (P) and diversity regularization (D). (c) Visualization of the distribution of candidate tasks (gray points) and selected tasks (red…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

100 extracted references · 37 canonical work pages

  1. [1]

    Sharp-maml: Sharpness-aware model-agnostic meta learning

    Abbas, M., Xiao, Q., Chen, L., Chen, P.-Y., and Chen, T. Sharp-maml: Sharpness-aware model-agnostic meta learning. In International conference on machine learning, pp.\ 10--32. PMLR, 2022

    2022

  2. [2]

    L., Nouri, A., and Wingate, D

    Asmuth, J., Li, L., Littman, M. L., Nouri, A., and Wingate, D. A bayesian sampling approach to exploration in reinforcement learning. arXiv preprint arXiv:1205.2664, 2012

    arXiv 2012

  3. [3]

    Finite-time analysis of the multiarmed bandit problem, 2002 a

    Auer, P. Finite-time analysis of the multiarmed bandit problem, 2002 a

    2002

  4. [4]

    Using confidence bounds for exploitation-exploration trade-offs

    Auer, P. Using confidence bounds for exploitation-exploration trade-offs. Journal of Machine Learning Research, 3 0 (Nov): 0 397--422, 2002 b

    2002

  5. [5]

    Z., Xiong, Z., Zintgraf, L., Finn, C., and Whiteson, S

    Beck, J., Vuorio, R., Liu, E. Z., Xiong, Z., Zintgraf, L., Finn, C., and Whiteson, S. A survey of meta-reinforcement learning. arXiv preprint arXiv:2301.08028, 2023

    arXiv 2023

  6. [6]

    C., and Ye, Y

    Borodin, A., Jain, A., Lee, H. C., and Ye, Y. Max-sum diversification, monotone submodular functions, and dynamic updates. ACM Transactions on Algorithms (TALG), 13 0 (3): 0 1--25, 2017

    2017

  7. [7]

    C., and Jordan, M

    Broderick, T., Boyd, N., Wibisono, A., Wilson, A. C., and Jordan, M. I. Streaming variational bayes. Advances in neural information processing systems, 26, 2013

    2013

  8. [8]

    On evaluating adversarial robustness

    Carlini, N., Athalye, A., Papernot, N., Brendel, W., Rauber, J., Tsipras, D., Goodfellow, I., Madry, A., and Kurakin, A. On evaluating adversarial robustness. arXiv preprint arXiv:1902.06705, 2019

    arXiv 1902

Show all 100 references
  1. [9]

    and Valko, M

    Carpentier, A. and Valko, M. Simple regret for infinitely many armed bandits. In International Conference on Machine Learning, pp.\ 1133--1141. PMLR, 2015

    2015

  2. [10]

    Risk aversion in finite markov decision processes using total cost criteria and average value at risk

    Carpin, S., Chow, Y.-L., and Pavone, M. Risk aversion in finite markov decision processes using total cost criteria and average value at risk. In 2016 ieee international conference on robotics and automation (icra), pp.\ 335--342. IEEE, 2016

    2016

  3. [11]

    Box2d: A 2d physics engine for games, 2007

    Catto, E. Box2d: A 2d physics engine for games, 2007. URL http://box2d.org

    2007

  4. [12]

    Tohan: A one-step approach towards few-shot hypothesis adaptation

    Chi, H., Liu, F., Yang, W., Lan, L., Liu, T., Han, B., Cheung, W., and Kwok, J. Tohan: A one-step approach towards few-shot hypothesis adaptation. Advances in Neural Information Processing Systems, 34: 0 20970--20982, 2021

    2021

  5. [13]

    Unveiling causal reasoning in large language models: Reality or mirage? Advances in Neural Information Processing Systems, 37: 0 96640--96670, 2024 a

    Chi, H., Li, H., Yang, W., Liu, F., Lan, L., Ren, X., Liu, T., and Han, B. Unveiling causal reasoning in large language models: Reality or mirage? Advances in Neural Information Processing Systems, 37: 0 96640--96670, 2024 a

    2024

  6. [14]

    Does confusion really hurt novel class discovery? International Journal of Computer Vision, 132 0 (8): 0 3191--3207, 2024 b

    Chi, H., Yang, W., Liu, F., Lan, L., Qin, T., and Han, B. Does confusion really hurt novel class discovery? International Journal of Computer Vision, 132 0 (8): 0 3191--3207, 2024 b

    2024

  7. [15]

    Risk-sensitive and data-driven sequential decision making

    Chow, Y. Risk-sensitive and data-driven sequential decision making. PhD thesis, Stanford University, 2017

    2017

  8. [16]

    Risk-sensitive and robust decision-making: a cvar optimization approach

    Chow, Y., Tamar, A., Mannor, S., and Pavone, M. Risk-sensitive and robust decision-making: a cvar optimization approach. Advances in neural information processing systems, 28, 2015

    2015

  9. [17]

    Risk-constrained reinforcement learning with percentile risk criteria

    Chow, Y., Ghavamzadeh, M., Janson, L., and Pavone, M. Risk-constrained reinforcement learning with percentile risk criteria. Journal of Machine Learning Research, 18 0 (167): 0 1--51, 2018

    2018

  10. [18]

    Safe policy learning for continuous control

    Chow, Y., Nachum, O., Faust, A., Due \ n ez-Guzman, E., and Ghavamzadeh, M. Safe policy learning for continuous control. In Conference on Robot Learning, pp.\ 801--821. PMLR, 2021

    2021

  11. [19]

    A., Ghahramani, Z., and Jordan, M

    Cohn, D. A., Ghahramani, Z., and Jordan, M. I. Active learning with statistical models. Journal of artificial intelligence research, 4: 0 129--145, 1996

    1996

  12. [20]

    Task-robust model-agnostic meta-learning

    Collins, L., Mokhtari, A., and Shakkottai, S. Task-robust model-agnostic meta-learning. Advances in Neural Information Processing Systems, 33: 0 18860--18871, 2020

    2020

  13. [21]

    Bullet physics simulation

    Coumans, E. Bullet physics simulation. In ACM SIGGRAPH 2015 Courses, pp.\ 1. 2015

    2015

  14. [22]

    Emergent complexity and zero-shot transfer via unsupervised environment design

    Dennis, M., Jaques, N., Vinitsky, E., Bayen, A., Russell, S., Critch, A., and Levine, S. Emergent complexity and zero-shot transfer via unsupervised environment design. Advances in neural information processing systems, 33: 0 13049--13061, 2020

    2020

  15. [23]

    L., Sutskever, I., and Abbeel, P

    Duan, Y., Schulman, J., Chen, X., Bartlett, P. L., Sutskever, I., and Abbeel, P. Rl2: Fast reinforcement learning via slow reinforcement learning. arXiv preprint arXiv:1611.02779, 2016

    2016 arXiv

  16. [24]

    Model-agnostic meta-learning for fast adaptation of deep networks

    Finn, C., Abbeel, P., and Levine, S. Model-agnostic meta-learning for fast adaptation of deep networks. In International conference on machine learning, pp.\ 1126--1135. PMLR, 2017

    2017

  17. [25]

    Probabilistic model-agnostic meta-learning

    Finn, C., Xu, K., and Levine, S. Probabilistic model-agnostic meta-learning. Advances in neural information processing systems, 31, 2018

    2018

  18. [26]

    Addressing function approximation error in actor-critic methods

    Fujimoto, S., Hoof, H., and Meger, D. Addressing function approximation error in actor-critic methods. In International conference on machine learning, pp.\ 1587--1596. PMLR, 2018

    2018

  19. [27]

    Deep bayesian active learning with image data

    Gal, Y., Islam, R., and Ghahramani, Z. Deep bayesian active learning with image data. In International conference on machine learning, pp.\ 1183--1192. PMLR, 2017

    2017

  20. [28]

    and Moulines, E

    Garivier, A. and Moulines, E. On upper-confidence bound policies for non-stationary bandit problems. arXiv preprint arXiv:0805.3415, 2008

    2008 arXiv

  21. [29]

    J., Eslami, S., and Teh, Y

    Garnelo, M., Schwarz, J., Rosenbaum, D., Viola, F., Rezende, D. J., Eslami, S., and Teh, Y. W. Neural processes. arXiv preprint arXiv:1807.01622, 2018

    2018 arXiv

  22. [30]

    W., Gast, J., Ruiz, I

    Gondal, M. W., Gast, J., Ruiz, I. A., Droste, R., Macri, T., Kumar, S., and Staudigl, L. Domain aligned clip for few-shot classification. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp.\ 5721--5730, 2024

    2024

  23. [31]

    Efficient risk-averse reinforcement learning

    Greenberg, I., Chow, Y., Ghavamzadeh, M., and Mannor, S. Efficient risk-averse reinforcement learning. Advances in Neural Information Processing Systems, 35: 0 32639--32652, 2022

    2022

  24. [32]

    Train hard, fight easy: Robust meta reinforcement learning

    Greenberg, I., Mannor, S., Chechik, G., and Meirom, E. Train hard, fight easy: Robust meta reinforcement learning. Advances in Neural Information Processing Systems, 36, 2024

    2024

  25. [33]

    Meta-reinforcement learning of structured exploration strategies

    Gupta, A., Mendonca, R., Liu, Y., Abbeel, P., and Levine, S. Meta-reinforcement learning of structured exploration strategies. Advances in neural information processing systems, 31, 2018

    2018

  26. [34]

    Beyond uniform sampling: Offline reinforcement learning with imbalanced datasets

    Hong, Z.-W., Kumar, A., Karnik, S., Bhandwaldar, A., Srivastava, A., Pajarinen, J., Laroche, R., Gupta, A., and Agrawal, P. Beyond uniform sampling: Offline reinforcement learning with imbalanced datasets. Advances in Neural Information Processing Systems, 36: 0 4985--5009, 2023

    2023

  27. [35]

    Meta-learning in neural networks: A survey

    Hospedales, T., Antoniou, A., Micaelli, P., and Storkey, A. Meta-learning in neural networks: A survey. IEEE transactions on pattern analysis and machine intelligence, 44 0 (9): 0 5149--5169, 2021

    2021

  28. [36]

    Replay-guided adversarial environment design

    Jiang, M., Dennis, M., Parker-Holder, J., Foerster, J., Grefenstette, E., and Rockt \"a schel, T. Replay-guided adversarial environment design. Advances in Neural Information Processing Systems, 34: 0 1884--1897, 2021

    2021

  29. [37]

    Kingma, D. P. and Welling, M. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, 2013

    2013 arXiv

  30. [38]

    Batchbald: Efficient and diverse batch acquisition for deep bayesian active learning

    Kirsch, A., Van Amersfoort, J., and Gal, Y. Batchbald: Efficient and diverse batch acquisition for deep bayesian active learning. Advances in neural information processing systems, 32, 2019

    2019

  31. [39]

    W., Sagawa, S., Marklund, H., Xie, S

    Koh, P. W., Sagawa, S., Marklund, H., Xie, S. M., Zhang, M., Balsubramani, A., Hu, W., Yasunaga, M., Phillips, R. L., Gao, I., et al. Wilds: A benchmark of in-the-wild distribution shifts. In International conference on machine learning, pp.\ 5637--5664. PMLR, 2021

    2021

  32. [40]

    D., Jansen, N., and Topcu, U

    Koprulu, C., Sim \ a o, T. D., Jansen, N., and Topcu, U. Risk-aware curriculum generation for heavy-tailed task distributions. In Uncertainty in Artificial Intelligence, pp.\ 1132--1142. PMLR, 2023

    2023

  33. [41]

    Offline reinforcement learning: Tutorial, review, and perspectives on open problems

    Levine, S., Kumar, A., Tucker, G., and Fu, J. Offline reinforcement learning: Tutorial, review, and perspectives on open problems. arXiv preprint arXiv:2005.01643, 2020

    2005 arXiv

  34. [42]

    Focal: Efficient fully-offline meta-reinforcement learning via distance metric learning and behavior regularization

    Li, L., Yang, R., and Luo, D. Focal: Efficient fully-offline meta-reinforcement learning via distance metric learning and behavior regularization. arXiv preprint arXiv:2010.01112, 2020

    2010 arXiv

  35. [43]

    Linsmeier, T. J. and Pearson, N. D. Value at risk. Financial analysts journal, 56 0 (2): 0 47--67, 2000

    2000

  36. [44]

    Theoretical investigations and practical enhancements on tail task risk minimization in meta learning

    Lv, Y., Wang, C., Liang, D., and Xie, Z. Theoretical investigations and practical enhancements on tail task risk minimization in meta learning. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. URL https://openreview.net/forum?id=McrzOo0hwr

    2024

  37. [45]

    J., Liang, W., Wang, H.-J., Wang, S., Zhu, Y., Fan, L., Bastani, O., and Jayaraman, D

    Ma, Y. J., Liang, W., Wang, H.-J., Wang, S., Zhu, Y., Fan, L., Bastani, O., and Jayaraman, D. Dreureka: Language model guided sim-to-real transfer. arXiv preprint arXiv:2406.01967, 2024

    2024 arXiv

  38. [46]

    and Teneketzis, D

    Mahajan, A. and Teneketzis, D. Multi-armed bandit problems. In Foundations and applications of sensor management, pp.\ 121--151. Springer, 2008

    2008

  39. [47]

    Supported value regularization for offline reinforcement learning

    Mao, Y., Zhang, H., Chen, C., Xu, Y., and Ji, X. Supported value regularization for offline reinforcement learning. Advances in Neural Information Processing Systems, 36: 0 40587--40609, 2023 a

    2023

  40. [48]

    Supported trust region optimization for offline reinforcement learning

    Mao, Y., Zhang, H., Chen, C., Xu, Y., and Ji, X. Supported trust region optimization for offline reinforcement learning. In International Conference on Machine Learning, pp.\ 23829--23851. PMLR, 2023 b

    2023

  41. [49]

    Offline reinforcement learning with ood state correction and ood action suppression

    Mao, Y., Wang, Q., Chen, C., Qu, Y., and Ji, X. Offline reinforcement learning with ood state correction and ood action suppression. arXiv preprint arXiv:2410.19400, 2024 a

    2024 arXiv

  42. [50]

    Doubly mild generalization for offline reinforcement learning

    Mao, Y., Wang, Q., Qu, Y., Jiang, Y., and Ji, X. Doubly mild generalization for offline reinforcement learning. arXiv preprint arXiv:2411.07934, 2024 b

    2024 arXiv

  43. [51]

    J., and Paull, L

    Mehta, B., Diaz, M., Golemo, F., Pal, C. J., and Paull, L. Active domain randomization. In Conference on Robot Learning, pp.\ 1162--1176. PMLR, 2020

    2020

  44. [52]

    H., and Gal, Y

    Mukhoti, J., Kirsch, A., van Amersfoort, J., Torr, P. H., and Gal, Y. Deep deterministic uncertainty: A new simple baseline. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 24384--24394, 2023

    2023

  45. [53]

    Domain randomization for simulation-based policy optimization with transferability assessment

    Muratore, F., Treede, F., Gienger, M., and Peters, J. Domain randomization for simulation-based policy optimization with transferability assessment. In Conference on Robot Learning, pp.\ 700--713. PMLR, 2018

    2018

  46. [54]

    Data-efficient domain randomization with bayesian optimization

    Muratore, F., Eilers, C., Gienger, M., and Peters, J. Data-efficient domain randomization with bayesian optimization. IEEE Robotics and Automation Letters, 6 0 (2): 0 911--918, 2021

    2021

  47. [55]

    V., Li, Y., Bui, T

    Nguyen, C. V., Li, Y., Bui, T. D., and Turner, R. E. Variational continual learning. arXiv preprint arXiv:1710.10628, 2017

    2017 arXiv

  48. [56]

    (more) efficient reinforcement learning via posterior sampling

    Osband, I., Russo, D., and Van Roy, B. (more) efficient reinforcement learning via posterior sampling. Advances in Neural Information Processing Systems, 26, 2013

    2013

  49. [57]

    Risk averse robust adversarial reinforcement learning

    Pan, X., Seita, D., Gao, Y., and Canny, J. Risk averse robust adversarial reinforcement learning. In 2019 International Conference on Robotics and Automation (ICRA), pp.\ 8522--8528. IEEE, 2019

    2019

  50. [58]

    Puterman, M. L. Markov decision processes: discrete stochastic dynamic programming. John Wiley & Sons, 2014

    2014

  51. [59]

    Meta-learning with neural bandit scheduler

    Qi, Y., Ban, Y., Wei, T., Zou, J., Yao, H., and He, J. Meta-learning with neural bandit scheduler. Advances in Neural Information Processing Systems, 36, 2024

    2024

  52. [60]

    Hokoff: Real game dataset from honor of kings and its offline reinforcement learning benchmarks

    Qu, Y., Wang, B., Shao, J., Jiang, Y., Chen, C., Ye, Z., Linc, L., Feng, Y., Lai, L., Qin, H., et al. Hokoff: Real game dataset from honor of kings and its offline reinforcement learning benchmarks. Advances in Neural Information Processing Systems, 36: 0 22166--22190, 2023

    2023

  53. [61]

    Choices are more important than efforts: Llm enables efficient multi-agent exploration

    Qu, Y., Wang, B., Jiang, Y., Shao, J., Mao, Y., Wang, C., Liu, C., and Ji, X. Choices are more important than efforts: Llm enables efficient multi-agent exploration. arXiv preprint arXiv:2410.02511, 2024

    2024 arXiv

  54. [62]

    Latent reward: Llm-empowered credit assignment in episodic reinforcement learning

    Qu, Y., Jiang, Y., Wang, B., Mao, Y., Wang, C., Liu, C., and Ji, X. Latent reward: Llm-empowered credit assignment in episodic reinforcement learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pp.\ 20095--20103, 2025

    2025

  55. [63]

    M., and Levine, S

    Rajeswaran, A., Finn, C., Kakade, S. M., and Levine, S. Meta-learning with implicit gradients. Advances in neural information processing systems, 32, 2019

    2019

  56. [64]

    Efficient off-policy meta-reinforcement learning via probabilistic context variables

    Rakelly, K., Zhou, A., Finn, C., Levine, S., and Quillen, D. Efficient off-policy meta-reinforcement learning via probabilistic context variables. In International conference on machine learning, pp.\ 5331--5340. PMLR, 2019

    2019

  57. [65]

    J., Fidler, S., and Litany, O

    Rempe, D., Philion, J., Guibas, L. J., Fidler, S., and Litany, O. Generating useful accident-prone driving scenarios via a learned traffic prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 17305--17315, 2022

    2022

  58. [66]

    B., Chen, X., and Wang, X

    Ren, P., Xiao, Y., Chang, X., Huang, P.-Y., Li, Z., Gupta, B. B., Chen, X., and Wang, X. A survey of deep active learning. ACM computing surveys (CSUR), 54 0 (9): 0 1--40, 2021

    2021

  59. [67]

    Risk-averse bayes-adaptive reinforcement learning

    Rigter, M., Lacerda, B., and Hawes, N. Risk-averse bayes-adaptive reinforcement learning. Advances in Neural Information Processing Systems, 34: 0 1142--1154, 2021

    2021

  60. [68]

    Been there, done that: Meta-learning with episodic recall

    Ritter, S., Wang, J., Kurth-Nelson, Z., Jayakumar, S., Blundell, C., Pascanu, R., and Botvinick, M. Been there, done that: Meta-learning with episodic recall. In International conference on machine learning, pp.\ 4354--4363. PMLR, 2018

    2018

  61. [69]

    T., Uryasev, S., et al

    Rockafellar, R. T., Uryasev, S., et al. Optimization of conditional value-at-risk. Journal of risk, 2: 0 21--42, 2000

    2000

  62. [70]

    and Van Roy, B

    Russo, D. and Van Roy, B. Learning to optimize via posterior sampling. Mathematics of Operations Research, 39 0 (4): 0 1221--1243, 2014

    2014

  63. [71]

    W., Hashimoto, T

    Sagawa, S., Koh, P. W., Hashimoto, T. B., and Liang, P. Distributionally robust neural networks. In International Conference on Learning Representations, 2019

    2019

  64. [72]

    Proximal policy optimization algorithms

    Schulman, J., Wolski, F., Dhariwal, P., Radford, A., and Klimov, O. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347, 2017

    2017 arXiv

  65. [73]

    Prompting is a double-edged sword: Improving worst-group robustness of foundation models

    Setlur, A., Garg, S., Smith, V., and Levine, S. Prompting is a double-edged sword: Improving worst-group robustness of foundation models. In Forty-first International Conference on Machine Learning, 2024

    2024

  66. [74]

    Counterfactual conservative q learning for offline multi-agent reinforcement learning

    Shao, J., Qu, Y., Chen, C., Zhang, H., and Ji, X. Counterfactual conservative q learning for offline multi-agent reinforcement learning. Advances in Neural Information Processing Systems, 36: 0 77290--77312, 2023 a

    2023

  67. [75]

    Complementary attention for multi-agent reinforcement learning

    Shao, J., Zhang, H., Qu, Y., Liu, C., He, S., Jiang, Y., and Ji, X. Complementary attention for multi-agent reinforcement learning. In International Conference on Machine Learning, pp.\ 30776--30793. PMLR, 2023 b

    2023

  68. [76]

    Sutton, R. S. and Barto, A. G. Reinforcement learning: An introduction. MIT press, 2018

    2018

  69. [77]

    Policy gradient for coherent risk measures

    Tamar, A., Chow, Y., Ghavamzadeh, M., and Mannor, S. Policy gradient for coherent risk measures. Advances in neural information processing systems, 28, 2015

    2015

  70. [78]

    Maniskill3: Gpu parallelized robotics simulation and rendering for generalizable embodied ai

    Tao, S., Xiang, F., Shukla, A., Qin, Y., Hinrichsen, X., Yuan, X., Bao, C., Lin, X., Liu, Y., Chan, T.-k., et al. Maniskill3: Gpu parallelized robotics simulation and rendering for generalizable embodied ai. arXiv preprint arXiv:2410.00425, 2024

    arXiv

  71. [79]

    Thompson, W. R. On the likelihood that one unknown probability exceeds another in view of the evidence of two samples. Biometrika, 25 0 (3-4): 0 285--294, 1933

    1933

  72. [80]

    Domain randomization via entropy maximization

    Tiboni, G., Klink, P., Peters, J., Tommasi, T., D'Eramo, C., and Chalvatzaki, G. Domain randomization via entropy maximization. arXiv preprint arXiv:2311.01885, 2023

    2023 arXiv

  73. [81]

    Domain randomization for transferring deep neural networks from simulation to the real world

    Tobin, J., Fong, R., Ray, A., Schneider, J., Zaremba, W., and Abbeel, P. Domain randomization for transferring deep neural networks from simulation to the real world. In 2017 IEEE/RSJ international conference on intelligent robots and systems (IROS), pp.\ 23--30. IEEE, 2017

    2017

  74. [82]

    Mujoco: A physics engine for model-based control

    Todorov, E., Erez, T., and Tassa, Y. Mujoco: A physics engine for model-based control. In 2012 IEEE/RSJ international conference on intelligent robots and systems, pp.\ 5026--5033. IEEE, 2012

    2012

  75. [83]

    N., Vapnik, V., et al

    Vapnik, V. N., Vapnik, V., et al. Statistical learning theory. 1998

    1998

  76. [84]

    Llm-empowered state representation for reinforcement learning

    Wang, B., Qu, Y., Jiang, Y., Shao, J., Liu, C., Yang, W., and Ji, X. Llm-empowered state representation for reinforcement learning. arXiv preprint arXiv:2407.13237, 2024 a

    2024 arXiv

  77. [85]

    Robust fast adaptation from adversarially explicit task distribution generation

    Wang, C., Lv, Y., Mao, Y., Qu, Y., Xu, Y., and Ji, X. Robust fast adaptation from adversarially explicit task distribution generation. arXiv preprint arXiv:2407.19523, 2024 b

    2024 arXiv

  78. [86]

    and Van Hoof, H

    Wang, Q. and Van Hoof, H. Doubly stochastic variational inference for neural processes with hierarchical latent variables. In International Conference on Machine Learning, pp.\ 10018--10028. PMLR, 2020

    2020

  79. [87]

    and Van Hoof, H

    Wang, Q. and Van Hoof, H. Learning expressive meta-representations with mixture of expert neural processes. Advances in neural information processing systems, 35: 0 26242--26255, 2022 a

    2022

  80. [88]

    and Van Hoof, H

    Wang, Q. and Van Hoof, H. Model-based meta reinforcement learning using graph structured surrogate models and amortized policy search. In International Conference on Machine Learning, pp.\ 23055--23077. PMLR, 2022 b

    2022

  81. [89]

    Bridge the inference gaps of neural processes via expectation maximization

    Wang, Q., Federici, M., and van Hoof, H. Bridge the inference gaps of neural processes via expectation maximization. In The Eleventh International Conference on Learning Representations, 2022

    2022

  82. [90]

    Bridge the inference gaps of neural processes via expectation maximization

    Wang, Q., Federici, M., and van Hoof, H. Bridge the inference gaps of neural processes via expectation maximization. In The Eleventh International Conference on Learning Representations, 2023 a

    2023

  83. [91]

    A simple yet effective strategy to robustify the meta learning paradigm

    Wang, Q., Lv, Y., Xie, Z., Huang, J., et al. A simple yet effective strategy to robustify the meta learning paradigm. Advances in Neural Information Processing Systems, 36, 2024 c

    2024

  84. [93]

    C., Xiao, Z., Mao, Y., Qu, Y., Shen, J., Lv, Y., and Ji, X

    Wang, Q. C., Xiao, Z., Mao, Y., Qu, Y., Shen, J., Lv, Y., and Ji, X. Model predictive task sampling for efficient and robust adaptation, 2025 b . URL https://arxiv.org/abs/2501.11039

    2025

  85. [94]

    Max-min diversification with fairness constraints: Exact and approximation algorithms

    Wang, Y., Mathioudakis, M., Li, J., and Fabbri, F. Max-min diversification with fairness constraints: Exact and approximation algorithms. In Proceedings of the 2023 SIAM International Conference on Data Mining (SDM), pp.\ 91--99. SIAM, 2023 b

    2023

  86. [95]

    Entropy-based active learning for object detection with progressive diversity constraint

    Wu, J., Chen, J., and Huang, D. Entropy-based active learning for object detection with progressive diversity constraint. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 9397--9406, 2022

    2022

  87. [96]

    Enhancing context-based meta-reinforcement learning algorithms via an efficient task encoder (student abstract)

    Xu, F., Jiang, S., Yin, H., Zhang, Z., Yu, Y., Li, M., Li, D., and Liu, W. Enhancing context-based meta-reinforcement learning algorithms via an efficient task encoder (student abstract). In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pp.\ 15937--...

    2021

  88. [97]

    Bayesian model-agnostic meta-learning

    Yoon, J., Kim, T., Dia, O., Kim, S., Bengio, Y., and Ahn, S. Bayesian model-agnostic meta-learning. Advances in neural information processing systems, 31, 2018

    2018

  89. [98]

    In-sample actor critic for offline reinforcement learning

    Zhang, H., Mao, Y., Wang, B., He, S., Xu, Y., and Ji, X. In-sample actor critic for offline reinforcement learning. In The Eleventh International Conference on Learning Representations, 2023

    2023

  90. [99]

    Combining active learning and semi-supervised learning using gaussian fields and harmonic functions

    Zhu, X., Lafferty, J., and Ghahramani, Z. Combining active learning and semi-supervised learning using gaussian fields and harmonic functions. In ICML 2003 workshop on the continuum from labeled to unlabeled data in machine learning and data mining, volume 3, pp.\ 58--65, 2003

    2003

  91. [100]

    Varibad: A very good method for bayes-adaptive deep rl via meta-learning

    Zintgraf, L., Shiarlis, K., Igl, M., Schulze, S., Gal, Y., Hofmann, K., and Whiteson, S. Varibad: A very good method for bayes-adaptive deep rl via meta-learning. arXiv preprint arXiv:1910.08348, 2019

    1910 arXiv

  92. [101]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.