Robust Bandwidth Estimation for Real-Time Communication with Offline Reinforcement Learning

Can Shen; Jian Kai; Tianwei Zhang; Yang Cao; Zihan Ling

REVIEW 4 major objections 3 minor 16 references

Robust Bandwidth Estimation for Real-Time Communication with Offline Reinforcement Learning

T0 review · 4 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Offline RL can be made safe for real-time bandwidth estimation by using a Q-ensemble and a Gaussian mixture policy, cutting overestimation by 18% and lifting worst-case QoE by 18.6%.

desk verdict Useful engineering integration with a credible deployment story, but the headline offline overestimation claim is under-specified and the sigma* definition looks wrong, so the paper needs revision before the results can be fully trusted. read the letter →

arxiv 2507.05785 v3 pith:UDTQJB7D submitted 2025-07-08 eess.SY cs.LGcs.SY

Jian Kai , Tianwei Zhang , Zihan Ling , Yang Cao , Can Shen This is my paper · ORCID

classification eess.SYcs.LGcs.SY

keywords bandwidthestimationreal-timecommunicationofflinereinforcementlearningQ-ensembleGaussianmixturepolicyout-of-distributiondetectionqualityofexperiencecongestioncontrol

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

RBWE is a two-stage framework for bandwidth estimation in real-time communication. It trains offline, without interacting with the live network, using implicit Q-learning on a large dataset of real RTC calls, and it represents the action policy as a Gaussian mixture so that the many different behavior policies in the dataset are captured rather than averaged away. During deployment it uses an ensemble of Q-functions to estimate uncertainty: when the ensemble agrees, it picks a conservative action from the policy's mode; when it disagrees, it falls back to the standard heuristic congestion controller. The paper reports that this design cuts the overestimation error rate by 18% relative to the behavior policies and raises the 10th-percentile QoE by 18.6% relative to the heuristic baseline, meaning the worst calls get better. If true, it shows a path to deploy learned rate control without risking the network disruptions that online exploration causes.

What carries the argument

The load-bearing machinery is the combination of a Q-ensemble with a Gaussian-mixture policy. The Q-ensemble, ten Q-networks sharing an architecture, supplies both a value estimate and an uncertainty signal: the expected-minimum approximation in Eq. (5) expresses the clipped Q-value as the ensemble mean minus a coefficient times the ensemble standard deviation, so overestimation is penalized explicitly. The Gaussian mixture policy matches the multimodal action distributions of heterogeneous behavior policies, avoiding the mode collapse of a single Gaussian. At deployment, the policy's local curvature at its mode, obtained from a second-order Taylor expansion of the log-density, defines a local standard deviation $\sigma^*$, and the lower-confidence-bound rule $a_{\text{chosen}} = a^* - \delta \sigma^*$ injects conservatism; the same Q-ensemble then computes the relative uncertainty $U_q(s,a)$ whose threshold triggers the fallback to the heuristic controller.

What would settle it

Re-run the online evaluation on the same roughly 400 call traces while sweeping $\delta$ from 0 to 1.5 and $\tau_u$ from 0.2 to 0.6; if RBWE's 10th-percentile QoE advantage over the heuristic collapses for a wide middle range of thresholds, or if a fixed conservative downscaling of the heuristic's estimate reproduces the same p10 QoE, then the claimed benefit is produced by the fallback bias rather than by the learned Q-ensemble policy.

Watch

Extended reading notes

Core claim

RBWE's central claim is that the obstacles to offline RL for bandwidth estimation—out-of-distribution actions, heterogeneous behavior policies, and deployment uncertainty—can be handled together by a single architecture. Offline training uses implicit Q-learning with an ensemble of ten Q-networks; the expected-minimum approximation turns the standard clipped-Q trick into an explicit mean-minus-variance penalty that suppresses overestimation. The policy is a four-component Gaussian mixture, so it can represent the multimodal action distributions left by different behavior policies instead of collapsing to one mode. At inference, the mode of the mixture is located by numerical optimization, a second-order Taylor expansion gives a local variance, and the chosen action is shifted down by $\delta$ times that local scale; if the ensemble's relative uncertainty $U_q(s,a)$ exceeds $\tau_u$, the system discards the learned action and keeps the heuristic bitrate. On the offline test set the overestimation error rate $e^+$ drops from 0.49 for the behavior policies to 0.27, and in online replay of about 400 emulated calls the 10th-percentile QoE reaches 60.0 versus 56.1 for the prior offline-RL baseline and 50.6 for the heuristic, which is the 18.6% improvement reported.

Load-bearing premise

The load-bearing premise is that the custom testbed's roughly 400 replayed emulated calls, together with the hand-set thresholds $\delta=0.5$ and $\tau_u=0.4$, represent real-world RTC conditions closely enough that the measured QoE gains are not artifacts of the replay setup.

Editorial extensions

If this is right

Offline-trained bandwidth policies can be deployed in production RTC without the exploration risk of online RL, because out-of-distribution actions are caught by Q-ensemble uncertainty and replaced by a heuristic bitrate.
Lower overestimation means fewer congestion-induced delay spikes and stalls, which is what moves the 10th-percentile QoE up by 18.6% relative to the heuristic baseline even though mean QoE gains are modest.
A Gaussian mixture policy makes a dataset collected under multiple production behavior policies usable as-is, without requiring the data to be filtered down to one policy.
The added safety machinery is cheap: about 1 ms inference after ONNX conversion, which fits comfortably inside the 60 ms control loop and can run on resource-constrained edge devices.
The tail improvement is the headline result: RBWE reaches a 10th-percentile QoE of 60.0 versus 56.1 for the prior offline-RL baseline and 50.6 for the heuristic, so the worst network conditions improve the most.

Reading between the lines

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

Editorial inference: a substantial part of the reported p10 QoE gain may come from the conservative bias and the fallback rather than from a better learned estimator; a fair comparison would pit RBWE against the heuristic with a constant downward bitrate bias.
Editorial inference: the fixed thresholds $\delta=0.5$ and $\tau_u=0.4$ are empirical, and the paper does not report sensitivity curves, so deployment-specific tuning could change the magnitude of the gains.
Editorial inference: the same pattern—an offline-learned policy plus an uncertainty-triggered fallback to a safe action—is transferable to other real-time control tasks such as adaptive bitrate streaming or datacenter congestion control, provided a trustworthy uncertainty signal and a safe fallback action exist.
Editorial inference: the offline evaluation reports only mean error rates, not the per-trace distribution of overestimation errors, so the stability claim would be stronger if the worst-case per-trace overestimation were reported as well.

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.

Desk Editor's Note

read the letter

The core idea here is sensible: combine IQL with a Q-ensemble and a Gaussian mixture policy, then add a conservative action-selection rule and an uncertainty-triggered fallback to GCC for deployment. That integration is new relative to Schaferct and Pioneer, and the deployment mechanism is a genuine attempt to solve a real problem: offline RL policies that look good offline can misbehave online. Credit where it's due: the paper ships a public implementation, evaluates on the Microsoft Teams dataset, and builds a real WebRTC testbed with about 400 emulated calls. That is formal, reproducible work, and the 10th percentile QoE improvement (60.0 vs 56.1 vs 50.6) is exactly the kind of tail metric that matters for RTC. The soft spots are real but not fatal. The most serious is the offline evaluation ambiguity: Table I reports overestimation rates, but the text never says whether the model output used is the raw mixture mode a* or the conservative LCB action a_chosen from Eq. (10). If it is a_chosen, then the 18% overestimation reduction is partly mechanical, a fixed downward bias of delta times sigma* will lower e+ by construction, and the paper provides no ablation separating the learned policy from the safety bias. The stress-test note is right about this. Relatedly, the definition sigma* = 1/lambda* in Section IV-B conflicts with the second-order expansion: for a Gaussian, the effective standard deviation should be 1/sqrt(lambda*), not 1/lambda*. That is a concrete error that changes the magnitude of the conservative correction and should be fixed or defended. Also: the deployment thresholds delta=0.5 and tau_u=0.4 are hand-set without sensitivity analysis, and the reward weight alpha is set once with no variation studied. The mean QoE gain over GCC (67.6 vs 65.9) is within one standard deviation; the tail improvement is the real signal. And it is odd to cite Pioneer [8] and then omit it from the benchmark table. None of this invalidates the central proposal. The paper is an honest engineering effort with better evidence than most in this subfield. The right fix is to clarify the offline evaluation protocol, run an ablation with delta=0, correct the sigma* definition, and add sensitivity plots. A serious referee could work through that in one round. Verdict: send to peer review. The paper is worth engaging with, and the open questions are answerable rather than structural. I would probably bring it to reading group too, mainly because the LCB-vs-learned-policy ambiguity is a nice teaching example of why evaluation protocols matter in offline RL.

Referee Report

4 major / 3 minor

Summary. The manuscript proposes RBWE, an offline reinforcement learning framework for bandwidth estimation (BWE) in real-time communication. It combines implicit Q-learning with an ensemble of Q-functions and a Gaussian mixture policy, trained on the Microsoft Teams BWE dataset. At deployment, the agent selects a lower-confidence-bound action based on the local shape of the mixture density and falls back to the GCC heuristic when Q-ensemble uncertainty is high. The paper reports offline results (Table I) that show a reduced overestimation error rate relative to behavior policies and Schaferct, and online testbed results (Table II) showing an improved 10th-percentile QoE relative to GCC and Schaferct.

Significance. If the empirical claims are supported, RBWE is a practically relevant contribution: it targets a real deployment problem, uses real-world session data, proposes a concrete mechanism for handling heterogeneous behavior policies and OOD actions, and provides a public implementation. The main strengths are the use of the Microsoft Teams dataset, the construction of a controlled WebRTC testbed, and the explicit fallback design for deployment stability. However, the central quantitative claims currently rest on an underspecified evaluation protocol and on hand-set deployment parameters whose sensitivity is not reported. The overestimation-reduction claim is not yet clearly separated from the effect of the conservative action-selection heuristic, and the online QoE gain is presented through a tail statistic without statistical support. These issues are fixable and do not undermine the overall approach, but they must be addressed before the reported improvements can be accepted as demonstrated.

major comments (4)

[§V-B, §IV-B] The offline evaluation protocol never states which action is used to produce Table I. If Table I is generated with a_chosen = a* − δσ* from Eq. (10) rather than the mode a* of Eq. (8), then the reduction in e+ from 0.49 (behavior policies) to 0.31 is expected by construction whenever σ* is positive, independent of the learned Q-values or the Gaussian mixture quality. Please specify the exact action-selection rule used in the offline evaluation, provide an ablation with δ = 0, and report a sensitivity sweep for δ. If the overestimation improvement disappears at δ = 0, the headline offline claim should be attributed to the safety heuristic rather than to the learned policy.
[§IV-B, Eq. (9)-(10)] The local Gaussian approximation is dimensionally inconsistent. The second-order Taylor expansion in Eq. (9) gives log π(a|s) ≈ log π(a*|s) − (1/2)λ*(a − a*)^2, so the effective standard deviation is σ* = 1/√λ*, not σ* = 1/λ* as stated immediately after Eq. (9). This changes the magnitude of the conservative shift δσ* in Eq. (10) and alters the interpretation of δ. The definition should be corrected before the LCB mechanism can be evaluated quantitatively.
[Abstract, §V-B, Table I] The '18%' overestimation-reduction claim is not well-defined. In Table I, e+ = 0.31 for RBWE versus 0.49 for behavior policies, which is an 18-percentage-point absolute difference and a 36.7% relative reduction; versus Schaferct (0.48) the difference is 17 percentage points. The abstract and Section V-B ('reduced by 17% and 18%') conflate percentage points and percent. Please report both absolute and relative reductions consistently, and state which comparison the abstract refers to.
[§V-C, Table II, Eq. (12)] The online QoE evidence for the headline 18.6% improvement is not statistically supported. The mean QoE values in Table II are 67.6 ± 5.6 (RBWE), 66.0 ± 5.5 (Schaferct), and 65.9 ± 10.3 (GCC), all within one standard deviation of each other; the 18.6% figure is the 10th-percentile difference from GCC. The paper does not report the number of calls contributing to the p10 estimate, confidence intervals, or a significance test, nor does it report the fraction of decisions in which the fallback to GCC (Section IV-B, τu = 0.4) is active. Reporting these would establish that the tail improvement is not an artifact of threshold choice or a small-sample tail statistic.

minor comments (3)

[§V-B] The explanation that 'mse is an absolute metric' and 'a slight underestimation can cause mse to soar' is plausible but not validated. Consider reporting a normalized error or a per-trace breakdown to show that the higher mse in Table I (3.49 versus 3.36) is not a systematic accuracy loss.
[§V-A] The hyperparameters N = 10, K = 4, τ = 0.7, β = 3, and α = 1.8 are reported without sensitivity analysis. Since α directly shapes the reward and δ and τu shape deployment behavior, a brief sensitivity study would materially strengthen the robustness claims.
[§VI] The conclusion contains a typo: 'Experimental evaluations had confirmed' should be 'Experimental evaluations have confirmed'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported gains are empirical measurements on held-out traces, and no central derived quantity reduces to its own inputs by construction.

full rationale

The paper's central claims are empirical rather than derivational. The offline overestimation reduction is measured against behavior policies and Schaferct on the 9,405 emulated challenge calls, while training uses a randomly selected subset of 1,800 calls per behavior policy, so the evaluation is held out from the training data. The Gaussian mixture policy and Q-ensemble are trained with standard IQL losses (Eqs. 2-4) using dataset rewards; no action, Q-value, or threshold is defined in terms of the reported e+ or QoE metrics. The conservative action selector (Eq. 10) and the OOD fallback (Eq. 11) are hand-set deployment parameters, not quantities fitted to the evaluation metrics, and the paper does not present them as predictions of the learned model. The QoE metric in Eq. (12) is externally specified in [5] and [16] and differs from the training reward, so the online QoE gain is not the training objective recycled as an evaluation. There is no load-bearing self-citation chain: the method builds on external, standard components (IQL [13], Q-ensemble [14], expected order statistics [15]) and the baselines are external. A residual ambiguity is that the offline evaluation does not explicitly state whether Table I uses the conservative action a_chosen or the mode a*, and a sensitivity analysis for delta would strengthen attribution of the overestimation reduction; this is an experimental-protocol caveat, not a circular derivation.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claim rests on several empirically chosen hyperparameters (alpha, delta, tau_u) and standard RL assumptions; no new physical entities are introduced. The main cost is the hand-tuning of deployment thresholds without sensitivity analysis.

free parameters (7)

alpha = 1.8
Weights audio and video MOS in the reward r = (2-alpha)*q_a + alpha*q_v; chosen because video is more sensitive to network changes.
delta = 0.5
Risk control parameter in the lower-confidence-bound action selection (10); set empirically, no sensitivity analysis.
tau_u = 0.4
Uncertainty threshold for OOD detection and fallback to GCC; set empirically, no sensitivity analysis.
K = 4
Number of Gaussian components in the mixture policy; chosen to capture multi-modal behavior policies.
N = 10
Number of Q-networks in the ensemble; larger N gives tighter uncertainty estimates but raises compute.
tau = 0.7
IQL expectile; controls pessimism in value estimation.
beta = 3.0
AWR temperature in policy extraction (4); balances Q-maximization against behavior cloning.

assumptions (5)

domain assumption Bandwidth estimation can be modeled as a finite-horizon MDP with the stated state, action, and reward
Section III defines the MDP; this assumes network and media features are Markovian enough for policy learning.
domain assumption Implicit Q-learning with expectile regression and AWR yields reliable offline policies
Section IV-A-b adopts IQL from [13]; correctness of IQL is taken as given.
domain assumption Q-values across the ensemble are Gaussian-distributed, justifying the expected-minimum approximation (5)
Equation (5) from [15] assumes normality of order statistics; the paper does not validate this on its Q-networks.
domain assumption The Microsoft Teams dataset is representative of real-world RTC network conditions
Section V-A uses this public dataset for offline training and emulated test calls; representativeness is assumed.
domain assumption The QoE metric in (12) captures user-perceived quality
Equation (12) combines network score and VMAF; the paper relies on its prior adoption [5].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Robust Bandwidth Estimation for Real-Time Communication with Offline Reinforcement Learning." pith.science (2026). https://pith.science/paper/UDTQJB7D

@misc{pith2026250705785,
  author       = {Pith},
  title        = {Pith review of: Robust Bandwidth Estimation for Real-Time Communication with Offline Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UDTQJB7D}},
  note         = {Machine review of arXiv:2507.05785}
}

read the original abstract

Accurate bandwidth estimation (BWE) is critical for real-time communication (RTC) systems. Traditional heuristic approaches offer limited adaptability under dynamic networks, while online reinforcement learning (RL) suffers from high exploration costs and potential service disruptions. Offline RL, which leverages high-quality data collected from real-world environments, offers a promising alternative. However, challenges such as out-of-distribution (OOD) actions, policy extraction from behaviorally diverse datasets, and reliable deployment in production systems remain unsolved. We propose RBWE, a robust bandwidth estimation framework based on offline RL that integrates Q-ensemble (an ensemble of Q-functions) with a Gaussian mixture policy to mitigate OOD risks and enhance policy learning. A fallback mechanism ensures deployment stability by switching to heuristic methods under high uncertainty. Experimental results show that RBWE reduces overestimation errors by 18% and improves the 10th percentile Quality of Experience (QoE) by 18.6%, demonstrating its practical effectiveness in real-world RTC applications. The implementation is publicly available at https://github.com/jiu2021/RBWE_offline.

Figures

Figures reproduced from arXiv: 2507.05785 by the authors.

**Figure 1.** , RBWE comprises two key stages: offline training and online deployment. Section IV-A provides a detailed description of the offline RL algorithm and its integration with the Gaussian mixture policy and Q-ensemble. Section IV-B elaborates on the online deployment strategy, which consists of two key components: conservative action estimation and OOD detection. A. Offline Training a) State, Action and Reward: We utili… view at source ↗

**Figure 2.** Output probability distribution of behavior policies. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Network architecture of Gaussian mixture policy. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: CDFs of Bandwidth Utilization, Network Score, Video Quality, and QoE. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Bandwidth estimation under four dynamic network traces. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 14 canonical work pages

[8]

Pioneer: Offline Reinforcement Learning based Bandwidth Estimation for Real- Time Communication,

B. Lu, K. Wang, J. Xu, R. Xie, L. Song, and W. Zhang, “Pioneer: Offline Reinforcement Learning based Bandwidth Estimation for Real- Time Communication,” inProceedings of the 15th ACM Multimedia Systems Conference, 2024, pp. 306–312

work page 2024
[1]

RFC 8825: Overview: Real-Time Protocols for Browser-based Applications,

H. Alvestrand, “RFC 8825: Overview: Real-Time Protocols for Browser-based Applications,” 2021

work page 2021
[2]

Analysis and Design of the Google Congestion Control for Web Real-Time Communication (WebRTC),

G. Carlucci, L. De Cicco, S. Holmer, and S. Mascolo, “Analysis and Design of the Google Congestion Control for Web Real-Time Communication (WebRTC),” inProceedings of the 7th International Conference on Multimedia Systems, 2016, pp. 1–12

work page 2016
[3]

From Ember to Blaze: Swift Interactive Video Adaptation via Meta- Reinforcement Learning,

X. Xiao, M. Yan, Y . Zuo, B. Liu, P. Ruan, Y . Cao, and W. Wang, “From Ember to Blaze: Swift Interactive Video Adaptation via Meta- Reinforcement Learning,” inIEEE INFOCOM 2023-IEEE Conference on Computer Communications. IEEE, 2023, pp. 1–10

work page 2023
[4]

OnRL: Improving Mobile Video Telephony via Online Rein- forcement Learning,

H. Zhang, A. Zhou, J. Lu, R. Ma, Y . Hu, C. Li, X. Zhang, H. Ma, and X. Chen, “OnRL: Improving Mobile Video Telephony via Online Rein- forcement Learning,” inProceedings of the 26th Annual International Conference on Mobile Computing and Networking, 2020, pp. 1–14

work page 2020
[5]

OpenNetLab: Open Platform for RL-based Congestion Control for Real-Time Communications,

J. Eo, Z. Niu, W. Cheng, F. Y . Yan, R. Gao, J. Kardhashi, S. Inglis, M. Revow, B.-G. Chun, P. Chenget al., “OpenNetLab: Open Platform for RL-based Congestion Control for Real-Time Communications,” in Proceedings of the 6th Asia-Pacific Workshop on Networking, 2022, pp. 70–75

work page 2022
[6]

ACM MMSys 2024 Bandwidth Estima- tion in Real Time Communications Challenge,

S. Khairy, G. Mittag, V . Gopal, F. Y . Yan, Z. Niu, E. Ameri, S. Inglis, M. Golestaneh, and R. Cutler, “ACM MMSys 2024 Bandwidth Estima- tion in Real Time Communications Challenge,” inProceedings of the 15th ACM Multimedia Systems Conference, 2024, pp. 339–345

work page 2024
[7]

Accurate Bandwidth Prediction for Real-Time Media Streaming with Offline Reinforcement Learning,

Q. Tan, G. Lv, X. Fang, J. Zhang, Z. Yang, Y . Jiang, and Q. Wu, “Accurate Bandwidth Prediction for Real-Time Media Streaming with Offline Reinforcement Learning,” inProceedings of the 15th ACM Multimedia Systems Conference, 2024, pp. 381–387

work page 2024

Show all 16 references

[9]

A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Prob- lems,

R. F. Prudencio, M. R. Maximo, and E. L. Colombini, “A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Prob- lems,”IEEE Transactions on Neural Networks and Learning Systems, 2023

2023
[10]

Mowgli: Passively Learned Rate Control for Real-Time Video,

N. Agarwal, R. Pan, F. Y . Yan, and R. Netravali, “Mowgli: Passively Learned Rate Control for Real-Time Video,” 2025. [Online]. Available: https://arxiv.org/abs/2410.03339

2025 arXiv
[11]

Learning to Coordinate Video Codec with Transport Protocol for Mobile Video Telephony,

A. Zhou, H. Zhang, G. Su, L. Wu, R. Ma, Z. Meng, X. Zhang, X. Xie, H. Ma, and X. Chen, “Learning to Coordinate Video Codec with Transport Protocol for Mobile Video Telephony,” inThe 25th Annual International Conference on Mobile Computing and Networking, 2019, pp. 1–16

2019
[12]

Loki: Improving Long Tail Performance of Learning-based Real-Time Video Adaptation by Fusing Rule-based Models,

H. Zhang, A. Zhou, Y . Hu, C. Li, G. Wang, X. Zhang, H. Ma, L. Wu, A. Chen, and C. Wu, “Loki: Improving Long Tail Performance of Learning-based Real-Time Video Adaptation by Fusing Rule-based Models,” inProceedings of the 27th Annual International Conference on Mobile Computin...

2021
[13]

Offline Reinforcement Learning with Implicit Q-Learning,

I. Kostrikov, A. Nair, and S. Levine, “Offline Reinforcement Learning with Implicit Q-Learning,” 2021. [Online]. Available: https://arxiv.org/abs/2110.06169

2021 arXiv
[14]

Uncertainty-based Offline Reinforcement Learning with Diversified Q-Ensemble,

G. An, S. Moon, J.-H. Kim, and H. O. Song, “Uncertainty-based Offline Reinforcement Learning with Diversified Q-Ensemble,”Advances in neural information processing systems, vol. 34, pp. 7436–7447, 2021

2021
[15]

Algorithm AS 177: Expected Normal Order Statistics(Exact and Approximate),

J. Royston, “Algorithm AS 177: Expected Normal Order Statistics(Exact and Approximate),”Applied Statistics, vol. 31, no. 2, pp. 161–5, 1982

1982
[16]

VMAF: The Netflix Video Multi-Method Assessment Fusion,

Netflix, “VMAF: The Netflix Video Multi-Method Assessment Fusion,” https://github.com/Netflix/vmaf, 2016, accessed: 2025-05-02

2016

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [8]

Pioneer: Offline Reinforcement Learning based Bandwidth Estimation for Real- Time Communication,

B. Lu, K. Wang, J. Xu, R. Xie, L. Song, and W. Zhang, “Pioneer: Offline Reinforcement Learning based Bandwidth Estimation for Real- Time Communication,” inProceedings of the 15th ACM Multimedia Systems Conference, 2024, pp. 306–312

work page 2024

[2] [1]

RFC 8825: Overview: Real-Time Protocols for Browser-based Applications,

H. Alvestrand, “RFC 8825: Overview: Real-Time Protocols for Browser-based Applications,” 2021

work page 2021

[3] [2]

Analysis and Design of the Google Congestion Control for Web Real-Time Communication (WebRTC),

G. Carlucci, L. De Cicco, S. Holmer, and S. Mascolo, “Analysis and Design of the Google Congestion Control for Web Real-Time Communication (WebRTC),” inProceedings of the 7th International Conference on Multimedia Systems, 2016, pp. 1–12

work page 2016

[4] [3]

From Ember to Blaze: Swift Interactive Video Adaptation via Meta- Reinforcement Learning,

X. Xiao, M. Yan, Y . Zuo, B. Liu, P. Ruan, Y . Cao, and W. Wang, “From Ember to Blaze: Swift Interactive Video Adaptation via Meta- Reinforcement Learning,” inIEEE INFOCOM 2023-IEEE Conference on Computer Communications. IEEE, 2023, pp. 1–10

work page 2023

[5] [4]

OnRL: Improving Mobile Video Telephony via Online Rein- forcement Learning,

H. Zhang, A. Zhou, J. Lu, R. Ma, Y . Hu, C. Li, X. Zhang, H. Ma, and X. Chen, “OnRL: Improving Mobile Video Telephony via Online Rein- forcement Learning,” inProceedings of the 26th Annual International Conference on Mobile Computing and Networking, 2020, pp. 1–14

work page 2020

[6] [5]

OpenNetLab: Open Platform for RL-based Congestion Control for Real-Time Communications,

J. Eo, Z. Niu, W. Cheng, F. Y . Yan, R. Gao, J. Kardhashi, S. Inglis, M. Revow, B.-G. Chun, P. Chenget al., “OpenNetLab: Open Platform for RL-based Congestion Control for Real-Time Communications,” in Proceedings of the 6th Asia-Pacific Workshop on Networking, 2022, pp. 70–75

work page 2022

[7] [6]

ACM MMSys 2024 Bandwidth Estima- tion in Real Time Communications Challenge,

S. Khairy, G. Mittag, V . Gopal, F. Y . Yan, Z. Niu, E. Ameri, S. Inglis, M. Golestaneh, and R. Cutler, “ACM MMSys 2024 Bandwidth Estima- tion in Real Time Communications Challenge,” inProceedings of the 15th ACM Multimedia Systems Conference, 2024, pp. 339–345

work page 2024

[8] [7]

Accurate Bandwidth Prediction for Real-Time Media Streaming with Offline Reinforcement Learning,

Q. Tan, G. Lv, X. Fang, J. Zhang, Z. Yang, Y . Jiang, and Q. Wu, “Accurate Bandwidth Prediction for Real-Time Media Streaming with Offline Reinforcement Learning,” inProceedings of the 15th ACM Multimedia Systems Conference, 2024, pp. 381–387

work page 2024