Pith. sign in

REVIEW 5 minor 40 references

Information Routing across Batch Boundaries: Memory--Batch Tradeoffs in Lipschitz Bandits

T0 review · 0 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper proves a near-tight minimax regret characterization for stochastic Lipschitz bandits with $W$-bit live memory and $B$ committed batches, introducing the joint penalty $T^{(d+2)/(d+3)}(1+(B-1)W)^{-1/(d(d+3))}$.

desk verdict A genuine joint memory-batch frontier for Lipschitz bandits, with a clean information-routing lower bound and careful matching constructions, but the model's no-uncharged-storage assumption and an overbroad static-vs-adaptive claim need qualification. read the letter →

arxiv 2608.07922 v1 pith:ICLMQMHS submitted 2026-08-08 cs.LG stat.ML

Zicheng Lyu , Zengfeng Huang This is my paper · ORCID
classification cs.LGstat.ML
keywords Lipschitzbanditsminimaxregretfinite-memorybatchedinformationroutingboundary-stateentropyactive-setmaskcommittedbatches
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

Adaptive learning needs both room to store what rewards imply and chances to act on that stored information. The paper proves that in stochastic Lipschitz bandits, a learner with at most $W$ bits of live reward-dependent state and at most $B$ committed batches suffers minimax expected pseudo-regret that, once $W \gtrsim_d \log(eT)$, is determined up to logarithmic factors by the worst of three terms: the classical sequential floor, the unrestricted-memory batch floor, and a new interaction involving $(B-1)W$. Lower bounds hold for every $W$, so the interaction is not confined to the memory-rich regime. The proof shows the interaction is an information-routing constraint: low regret forces the committed action transcript to encode a growing number of regional decisions, while the boundary states that can influence later experiments carry at most $(B-1)W$ bits in total. This matters because it says memory width and update depth are complementary resources, not substitutes: one wide state used by very few redesigns cannot replace a sequence of narrow states recomputed between rounds of exploration.

What carries the argument

The carrying object is the boundary-state entropy budget $\chi=(B-1)W$. Lemma 3.2 shows that, conditional on the public seed, every reward-dependent element of the committed action transcript is a measurable function of the tuple of nonterminal boundary states, so for any latent instance variable $V$ the information profile satisfies $I(V;\mathcal{T}\mid F_0) \le H(M\mid F_0) \le \chi$. Against this budget, the hard family places $m \asymp_d s^{-d}$ separated pairs of scale-$s$ regions with $q \asymp_d (s/r)^d$ radius-$r$ probes per region; a stopped change-of-measure argument forces any low-regret transcript to encode $\Theta_d(s^{-d})$ regional decisions, and Fano decoding then yields the resolution floor $s_{\mathrm{mem}}=(1+\chi)^{-1/d}$. The upper bound is a safe active-set mask of the same spatial order, maintained either entirely in memory or as regenerated one-memory-sized fragments, while all fine verification statistics are accumulated, compared, and erased.

What would settle it

Simulate the $d=1$ hard family with $B=2$ and $W=0$ over horizons $T=10^2,\dots,10^6$: the theorem says the best 0-bit 2-batch policy has linear regret, so any schedule whose normalized regret decays with $T$ would refute the lower bound.

Watch

Extended reading notes

Core claim

The central claim is Theorem 3.4: for $d \ge 1$, $T \ge T_d$, $B \in [T]$, and $W \ge 0$, the minimax expected pseudo-regret satisfies $R_T(B,W) \ge c_d [\Psi_T(s_{T,\chi}) \vee T^{\beta_{d,B}}/B^2]$, and whenever $W \ge C_d \log(eT)$ it also satisfies $R_T(B,W) \le C_d \log(eT)[\Psi_T(s_{T,\chi}) \vee T^{\beta_{d,B}}/B^2]$, with $\alpha_d=(d+1)/(d+2)$, $\beta_{d,B}=\alpha_d/(1-(d+2)^{-B})$, $\chi=(B-1)W$, and $s_{T,\chi}=\max\{T^{-1/(d+2)}, (1+\chi)^{-1/d}\}$, where $\Psi_T(s)=T^{(d+2)/(d+3)}s^{1/(d+3)}$. Explicitly, the frontier is the maximum of $T^{\alpha_d}$, $T^{\beta_{d,B}}/B^2$, and the new interaction $T^{(d+2)/(d+3)}(1+(B-1)W)^{-1/(d(d+3))}$. The same regional hard family produces the sequential and interaction terms, and an adaptive-grid transfer produces the update-depth term; matching policies use static batch boundaries and retain only an active-set mask while streaming and erasing verification statistics. Thus the result recovers the batch-only frontier with unrestricted memory and the fully sequential minimax rate with logarithmic live memory as specializations.

Load-bearing premise

The lower bound rests on the assumption that the declared $W$-bit state is the only reward-dependent storage, with all randomness and schedules fixed by the public seed; if reward-dependent information can persist outside that state—in hidden registers, external transcripts, or reward-dependent timing not captured by the state—the entropy budget $(B-1)W$ and the entire routing lower bound would no longer hold.

Editorial extensions

If this is right

  • Once $W \ge C_d \log(eT)$, the joint memory–batch minimax regret is characterized up to logarithmic factors, and static batch boundaries achieve the worst-case optimal order.
  • Near-sequential regret requires both $B \gtrsim \log\log T$ and $(B-1)W \gtrsim T^{d/(d+2)}$; the batch complexity is $\widetilde{\Theta}_d(\log\log T \vee T^{d/(d+2)}/W)$.
  • With unrestricted memory, the theorem recovers the full-dimensional batch-only frontier $T^{\alpha_d} \vee T^{\beta_{d,B}}/B^2$ for every batch budget $B$.
  • With $B=T$, the model is fully sequential and the theorem recovers the classical minimax regret $\widetilde{\Theta}_d(T^{(d+1)/(d+2)})$ using only logarithmic live memory.
  • The new interaction penalty shows concretely that state width and update depth are not interchangeable: concentrating the same boundary-state entropy into fewer, wider states cannot reproduce the sequence of refinements needed.

Reading between the lines

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

  • Beyond the paper's terminal regret bounds, the prefix profile $I(V;\mathcal{T}[j]\mid F_0) \le (j-1)W$ suggests a time-resolved converse: the same Markov factorization should yield per-batch regret lower bounds, not just an end-of-horizon statement.
  • The mechanism is likely to transfer to other nonparametric classes where the number of relevant regions grows with resolution: the effective dimension in the memory floor may become the zooming or near-optimality dimension of the class, not the ambient dimension.
  • The serialized active-set construction is a concrete streaming template: regenerate a memory-sized mask fragment, commit the corresponding child probes, update one resident best record, and erase the fragment; the same pattern could be exported to memory-bounded hierarchical optimization on devices with no random-access storage.
  • A natural testable next step is the sublogarithmic regime $W<\log(eT)$: the very-low-entropy codebook bound $R_T(B,W) \ge c_d T2^{-\chi/d}$ suggests an exponential-in-$\chi$ frontier there, but the paper leaves this regime open.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 5 minor

Summary. This paper studies a joint resource model for stochastic Lipschitz bandits on [0,1]^d: after each pull the learner may retain at most W bits of live reward-dependent state, and its pulls are organized into at most B committed batches whose boundaries and action tapes are fixed at batch boundaries. The main result, Theorem 3.4, gives a two-sided minimax characterization up to logarithmic factors when W is at least a dimension-dependent constant times log(eT), with a lower bound valid for every W. The regret frontier is the maximum of three terms: the classical sequential term T^{alpha_d}, a batch-depth term T^{beta_{d,B}}/B^2, and a new interaction penalty T^{(d+2)/(d+3)}(1+(B-1)W)^{-1/(d(d+3))}. The lower bound is proved through a regional hard family in which low regret forces the committed action transcript to encode many regional routing decisions, while the boundary states carry at most (B-1)W bits of entropy; a separate adaptive-grid argument supplies the batch-depth branch. The upper bound maintains only a safe active-set mask, either in memory or in regenerated fragments, while streaming and erasing verification statistics. Corollaries recover the unrestricted-memory batched frontier and the fully sequential logarithmic-memory specialization, and static batch boundaries are shown to match predictable adaptive ones in the characterized regime.

Significance. If the result is correct, it is a substantive contribution to the theory of resource-constrained bandits. The paper identifies a genuinely new interaction term coupling state width W and batch depth B, and it proves that the two resources are not interchangeable, which is the paper's central conceptual claim. The lower and upper bounds are matched up to logarithmic factors, the theorem cleanly specializes to the previously known batch-only and fully sequential finite-memory frontiers, and the appendices give a detailed, largely self-contained accounting of the geometry, the information-theoretic boundary-state argument, and the streaming algorithms. The model is carefully defined, including the crucial convention that there is no uncharged reward-dependent workspace or accumulating external transcript; this convention is load-bearing for the lower bound, but it is stated explicitly rather than hidden. The main limitations are the logarithmic gaps, the restriction of the matching upper bound to W >= C_d log(eT), and the fact that the sublogarithmic regime is left open.

minor comments (5)
  1. [Section 3.1 and Appendix A.1] The model description should be tightened around the status of the committed action tape. The text first says there is no separate persistent reward-dependent workspace or accumulating external transcript, but then allows a fixed read-only action tape that replays information already encoded in the preceding W-bit boundary state. Appendix A.1's formal map-level definition has no tape variable, yet the upper-bound memory accounting in Proposition 5.3 and Appendix D implicitly treats the committed tape as free. Please state explicitly that W is the read-write workspace, while the committed tape is an uncounted read-only output determined by the boundary state; this would remove an apparent contradiction and make the lower-bound convention unambiguous.
  2. [Sections 1 and 2] The statements 'Predictable adaptive boundaries do not improve the worst-case order over static ones' are proved only in the regime W >= C_d log(eT), since the matching static upper bound in Theorem 3.4 has exactly that hypothesis and the sublogarithmic regime is declared open in Section 6. Please qualify these claims by adding 'for W >= C_d log(eT)' or by placing them explicitly under the theorem's stated condition.
  3. [Remark 3.5] In the d=1 specialization, the condition for near-sequential regret is stated as 'chi >= T^{1/3}', but the exact condition derived from the (1+chi)^{-1/(d(d+3))} term is (1+chi) >= c T^{1/3}. Consider writing '1+chi >= T^{1/3}' (with a dimension constant) to match Eq. (3) exactly.
  4. [Corollary 3.8 and Appendix F] The displayed lower bound in Eq. (6) contains a factor of ell_T^{-d(d+3)}, while the upper bound has no ell_T factor. Since Eq. (7) then suppresses logarithmic factors, it would be helpful to add a sentence stating explicitly that Eq. (6) does not track logarithmic factors, or to include the exact log exponents in both bounds so the displayed asymmetry is intentional.
  5. [Section 3.2] In Lemma 3.3, the Markov chain V -> M -> T -> R is stated conditionally on F0. The independence of the decoder randomization from the experiment makes the final step valid, but a one-sentence reminder after the lemma would help readers see why R is conditionally independent of V given T rather than merely generated in a complicated way from the transcript.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the interaction term is derived from an explicit hard family and boundary-state entropy budget, not from a fitted parameter or self-citation chain.

full rationale

The central new quantity, the width–depth interaction involving (B−1)W, is derived self-containedly. Section 3.2 defines the boundary-state tuple M and proves the reconstruction and entropy lemmas (Lemmas 3.2 and 3.3), giving I(V;T|F0) ≤ H(M|F0) ≤ (B−1)W by direct counting of W-bit boundary states. Section 4.2 then constructs a separated regional hard family, uses a stopped change-of-measure argument (Lemma B.4, proved in Appendix B.3) and Fano-style decoding to show low regret forces Ω(s^{-d}) bits of routing information, yielding the s ≳ (1+(B−1)W)^{-1/d} floor. The upper bound in Section 5 realizes the matching active-set mask with the same scale balance. No parameter is fitted to a subset of the regret data and then renamed a prediction. The batch-depth branch T^{β_d,B}/B^2 is imported from Feng et al. (2024), a prior peer-reviewed external result, and the stopped-comparison technique of Huang et al. (2026) is reproved in the appendix; these are independent support, not restatements of the present theorem. The explicit modeling assumption in Section 3.1 that there is no separate reward-dependent workspace or accumulating external transcript is a stated interface condition, not a circular restatement of the conclusion. Section 6 openly declares the sublogarithmic-memory regime open, so the Section 1 statement that predictable adaptive boundaries do not improve over static ones should be read under the theorem's W≳log(eT) hypothesis; this is a scope caveat, not a circular step.

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

The frontier is derived from the concrete resource interface rather than from fitted numeric constants. All dimension-dependent constants are existential proof constants with no scientific value fitted to data. The only imported ingredient is the prior adaptive-grid lower bound for the batch-depth branch. No new physical or empirical entities are postulated; 'boundary-state entropy budget', 'active-set mask', and 'regional routing vector' are proof objects, not independently falsifiable entities.

assumptions (4)
  • domain assumption Rewards are conditionally independent draws from Borel kernels with mean f(A_t), where f is one-Lipschitz on [0,1]^d with values in [0,1].
    Stated in Section 3.1; the whole minimax analysis is relative to this stochastic Lipschitz class.
  • domain assumption The learner's complete mutable reward-dependent state is exactly W bits and there is no separate reward-dependent workspace or accumulating transcript; all geometric schedules are fixed conditional on public seed F0.
    Section 3.1 and Appendix A.1. If hidden reward-dependent storage existed, the H(M|F0) at most (B-1)W bound in Lemma 3.3 and the codebook range bound in Lemma 3.2 would fail.
  • standard math The batch-depth term of the lower bound imports the adaptive-grid lower bound of Feng et al. (2024) and transfers it from Gaussian to bounded Bernoulli rewards.
    Used for the T^{beta_{d,B}}/B^2 branch; the transfer argument is supplied in Lemmas B.6 and B.7, but the underlying hard family is not re-derived.
  • standard math Standard information inequalities including conditional data processing, Pinsker, Fano, and the stopped change-of-measure lemma hold as stated.
    Background used throughout Section 4; Lemma B.4 is proved in the appendix, while the classical inequalities are taken as standard.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Information Routing across Batch Boundaries: Memory--Batch Tradeoffs in Lipschitz Bandits." pith.science (2026). https://pith.science/paper/ICLMQMHS

@misc{pith2026260807922,
  author       = {Pith},
  title        = {Pith review of: Information Routing across Batch Boundaries: Memory--Batch Tradeoffs in Lipschitz Bandits},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ICLMQMHS}},
  note         = {Machine review of arXiv:2608.07922}
}
abstract

Adaptive learning needs both a state that preserves what observations imply and opportunities to act on that state. We study this width--depth tradeoff in stochastic Lipschitz bandits. After each pull, the learner retains at most $W$ bits of live reward-dependent state and organizes its pulls into at most $B$ committed batches. For $W\gtrsim_d\log(eT)$, we characterize minimax expected pseudo-regret up to logarithmic factors; the lower bounds hold for every $W$. Besides the classical sequential and unrestricted-memory batch costs, the frontier contains the new penalty \[ T^{\frac{d+2}{d+3}} \bigl(1+(B-1)W\bigr)^{-\frac1{d(d+3)}}, \] proving that state width and update depth are not interchangeable. The interaction is an information-routing constraint: at regional scale $s$, low regret forces the committed action transcript to encode $\Theta_d(s^{-d})$ regional decisions, while the collected boundary states carry at most $(B-1)W$ bits of entropy. Matching policies stream and erase verification statistics while retaining a mask of a safe active set, either in memory or fragment by fragment. The theorem recovers the full-dimensional worst-case batch-only frontier and logarithmic-memory achievability in the fully sequential specialization; static batch boundaries match predictable adaptive ones.

Figures

Figures reproduced from arXiv: 2608.07922 by the authors.

Figure 1
Figure 1. One hard family exposes both resource budgets. Every scale- [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 38 canonical work pages

  1. [1]

    SIAM Journal on Control and Optimization , volume=

    The continuum-armed bandit problem , author=. SIAM Journal on Control and Optimization , volume=. 1995 , doi=

    work page 1995

  2. [2]

    Advances in Neural Information Processing Systems , year=

    Nearly tight bounds for the continuum-armed bandit problem , author=. Advances in Neural Information Processing Systems , year=

    work page

  3. [3]

    Proceedings of the 20th Conference on Learning Theory , pages=

    Improved rates for the stochastic continuum-armed bandit problem , author=. Proceedings of the 20th Conference on Learning Theory , pages=

    work page

  4. [4]

    Proceedings of the Fortieth Annual ACM Symposium on Theory of Computing , pages=

    Multi-armed bandits in metric spaces , author=. Proceedings of the Fortieth Annual ACM Symposium on Theory of Computing , pages=. 2008 , doi=

    work page 2008

  5. [5]

    Proceedings of the Twenty-Third Conference on Uncertainty in Artificial Intelligence , pages=

    Bandit algorithms for tree search , author=. Proceedings of the Twenty-Third Conference on Uncertainty in Artificial Intelligence , pages=

    work page

  6. [6]

    Bubeck, Sebastien and Munos, Remi and Stoltz, Gilles and Szepesvari, Csaba , journal=

    work page

  7. [7]

    Advances in Neural Information Processing Systems , year=

    Optimistic optimization of a deterministic function without the knowledge of its smoothness , author=. Advances in Neural Information Processing Systems , year=

    work page

  8. [8]

    Proceedings of the 27th Conference on Learning Theory , pages=

    Lipschitz Bandits: Regret Lower Bound and Optimal Algorithms , author=. Proceedings of the 27th Conference on Learning Theory , pages=. 2014 , volume=

    work page 2014

Show all 40 references
  1. [9]

    Proceedings of the 34th Conference on Learning Theory , pages=

    Adaptive discretization for adversarial Lipschitz bandits , author=. Proceedings of the 34th Conference on Learning Theory , pages=

  2. [10]

    The Annals of Statistics , volume=

    Batched bandit problems , author=. The Annals of Statistics , volume=. 2016 , doi=

    2016

  3. [11]

    Advances in Neural Information Processing Systems , volume=

    Batched multi-armed bandits problem , author=. Advances in Neural Information Processing Systems , volume=

  4. [12]

    Proceedings of the 38th International Conference on Machine Learning , pages=

    Almost optimal anytime algorithm for batched multi-armed bandits , author=. Proceedings of the 38th International Conference on Machine Learning , pages=

  5. [13]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Regret bounds for batched bandits , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=. 2021 , doi=

    2021

  6. [14]

    Proceedings of the 53rd Annual ACM SIGACT Symposium on Theory of Computing , pages=

    Linear bandits with limited adaptivity and learning distributional optimal design , author=. Proceedings of the 53rd Annual ACM SIGACT Symposium on Theory of Computing , pages=. 2021 , doi=

    2021

  7. [15]

    IEEE Transactions on Information Theory , volume=

    Lipschitz Bandits With Batched Feedback , author=. IEEE Transactions on Information Theory , volume=. 2024 , doi=. 2110.09722 , archivePrefix=

    2024 arXiv

  8. [16]

    The Thirteenth International Conference on Learning Representations , year=

    Lipschitz Bandits in Optimal Space , author=. The Thirteenth International Conference on Learning Representations , year=

  9. [17]

    Information and Control , volume=

    A note on the two-armed bandit problem with finite memory , author=. Information and Control , volume=. 1968 , doi=

    1968

  10. [18]

    IEEE Transactions on Information Theory , volume=

    The two-armed-bandit problem with time-invariant finite memory , author=. IEEE Transactions on Information Theory , volume=. 1970 , doi=

    1970

  11. [19]

    Proceedings of the 21st International Conference on Artificial Intelligence and Statistics , pages=

    Stochastic multi-armed bandits in constant space , author=. Proceedings of the 21st International Conference on Artificial Intelligence and Statistics , pages=

  12. [20]

    Proceedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing , pages=

    Exploration with limited memory: Streaming algorithms for coin tossing, noisy comparisons, and multi-armed bandits , author=. Proceedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing , pages=. 2020 , doi=

    2020

  13. [21]

    Proceedings of the 35th Conference on Learning Theory , pages=

    A Sharp Memory-Regret Trade-off for Multi-Pass Streaming Bandits , author=. Proceedings of the 35th Conference on Learning Theory , pages=. 2022 , volume=

    2022

  14. [22]

    Proceedings of the 37th Conference on Learning Theory , pages=

    The best arm evades: Near-optimal multi-pass streaming lower bounds for pure exploration in multi-armed bandits , author=. Proceedings of the 37th Conference on Learning Theory , pages=

  15. [23]

    Proceedings of the 57th IEEE Symposium on Foundations of Computer Science , pages=

    Fast learning requires good memory: A time-space lower bound for parity learning , author=. Proceedings of the 57th IEEE Symposium on Foundations of Computer Science , pages=. 2016 , doi=

    2016

  16. [24]

    Proceedings of the 50th Annual ACM SIGACT Symposium on Theory of Computing , pages=

    Extractor-Based Time-Space Lower Bounds for Learning , author=. Proceedings of the 50th Annual ACM SIGACT Symposium on Theory of Computing , pages=. 2018 , publisher=

    2018

  17. [25]

    arXiv preprint arXiv:2603.13742 , year=

    Few batches or little memory, but not both: Simultaneous space and adaptivity constraints in stochastic bandits , author=. arXiv preprint arXiv:2603.13742 , year=. 2603.13742 , archivePrefix=

  18. [26]

    Elements of Information Theory , author=

  19. [27]

    Advances in Neural Information Processing Systems , volume=

    Generalized linear bandits with limited adaptivity , author=. Advances in Neural Information Processing Systems , volume=

  20. [28]

    Advances in Neural Information Processing Systems , volume=

    Efficient batched algorithm for contextual linear bandits with large action space via soft elimination , author=. Advances in Neural Information Processing Systems , volume=

  21. [29]

    arXiv preprint arXiv:2306.07903 , year=

    Tight memory-regret lower bounds for streaming bandits , author=. arXiv preprint arXiv:2306.07903 , year=. 2306.07903 , archivePrefix=

    arXiv

  22. [30]

    Proceedings of the 40th International Conference on Machine Learning , pages=

    Tight regret bounds for single-pass streaming multi-armed bandits , author=. Proceedings of the 40th International Conference on Machine Learning , pages=

  23. [31]

    Proceedings of the 30th Conference on Learning Theory , pages=

    Learning with limited rounds of adaptivity: Coin tossing, multi-armed bandits, and ranking from pairwise comparisons , author=. Proceedings of the 30th Conference on Learning Theory , pages=

  24. [32]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Regret Minimisation in Multi-Armed Bandits using Bounded Arm Memory , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=. 2020 , doi=

    2020

  25. [33]

    Advances in Neural Information Processing Systems , volume=

    Multi-Armed Bandits with Bounded Arm-Memory: Near-Optimal Guarantees for Best-Arm Identification and Regret Minimization , author=. Advances in Neural Information Processing Systems , volume=

  26. [34]

    Proceedings of the 38th International Conference on Machine Learning , pages=

    Optimal streaming algorithms for multi-armed bandits , author=. Proceedings of the 38th International Conference on Machine Learning , pages=

  27. [35]

    Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing , pages=

    Memory Bounds for the Experts Problem , author=. Proceedings of the 54th Annual ACM SIGACT Symposium on Theory of Computing , pages=. 2022 , publisher=

    2022

  28. [36]

    Proceedings of the 2023 Annual ACM-SIAM Symposium on Discrete Algorithms , pages=

    Online Prediction in Sub-linear Space , author=. Proceedings of the 2023 Annual ACM-SIAM Symposium on Discrete Algorithms , pages=. 2023 , publisher=

    2023

  29. [37]

    Proceedings of the 64th IEEE Annual Symposium on Foundations of Computer Science , pages=

    Near Optimal Memory-Regret Tradeoff for Online Learning , author=. Proceedings of the 64th IEEE Annual Symposium on Foundations of Computer Science , pages=. 2023 , publisher=

    2023

  30. [38]

    2024 , note=

    Efficient Algorithms for Lipschitz Bandits , author=. 2024 , note=

    2024

  31. [39]

    International Conference on Learning Representations , year=

    On Bits and Bandits: Quantifying the Regret--Information Trade-off , author=. International Conference on Learning Representations , year=. 2405.16581 , archivePrefix=

    arXiv

  32. [40]

    Batched Stochastic Linear Bandits with

    Lau, Ivan and McMorrow, Daniel and Jamieson, Kevin and Scarlett, Jonathan , journal=. Batched Stochastic Linear Bandits with. 2026 , eprint=

    2026

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.