Pith. sign in

REVIEW 5 major objections 5 minor 8 cited by

The paper claims that diffusion-based auto-bidding can be made reliable at scale by training a "completer" to finish observed trajectories and an "aligner" to refine the completion toward the advertiser's objective, yielding a 29.9% value g

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-05 10:58 UTC pith:BKB4L5OW

load-bearing objection A useful industrial auto-bidding result with a solid but not fully isolated mechanism; the completer's causal story needs an ablation before the 29.9% headline survives scrutiny. the 5 major comments →

arxiv 2509.03348 v1 pith:BKB4L5OW submitted 2025-09-03 cs.GT cs.LG

Generative Auto-Bidding in Large-Scale Competitive Auctions via Diffusion Completer-Aligner

Yewen Li , Jingtong Gao , Nan Jiang , Shuai Mao , Ruyi An , Fei Pan , Xiangyu Zhao , Bo An
show 2 more authors
Qingpeng Cai Peng Jiang
This is my paper
classification cs.GT cs.LG MSC 91B2668T07
keywords Auto-biddingDiffusion modelsComputational advertisingDynamic legitimacyTrajectory completionOffline reinforcement learningAuctionNetSparse rewards
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper argues that diffusion-based auto-bidding failed at scale not because trajectories are hard to generate, but because generated states are dynamically illegitimate—the planned next state cannot actually follow from the observed one—and because sampling noise drifts away from the advertiser's target. It proposes to fix both: train a "completer" by masking a random prefix of the trajectory and denoising only the masked future part, then, at inference, refine the completed future states with a gradient step from a trajectory-level return model, the "aligner." On the AuctionNet benchmark, the full method raises conversion value by up to 29.9% in the sparse-reward setting, and the deployed system reports a 2.0% gain in target cost on Kuaishou's advertising platform. If the causal story holds, the completer loss is the larger half of the gain: ablations show it alone outperforms DiffBid, while transformer and causal-attention alternatives do not.

Core claim

The paper's central claim is that generation uncertainty—not model capacity or reward sparsity—is the binding constraint on diffusion-based auto-bidding in large-scale competitive auctions, and that this uncertainty decomposes into two tractable problems. The first is dynamic legitimacy: a diffuser trained only to denoise whole trajectories has no incentive to make adjacent states causally consistent, so the paper augments the training objective with a random truncation t and a completion loss Lc that scores the model only on the future positions t+1:T while keeping the observed query s0:t clean; this teaches the model the conditional distribution p(st+1:T | s0:t, y(τ)) in a single forward p

What carries the argument

The completer loss (Eq. 11) is a masked denoising objective: the time index t is sampled uniformly, the observed prefix s0:t is injected clean into every noisy trajectory, and the squared noise-prediction error is computed only on positions t+1:T. This forces a single diffusion model to learn next-state consistency at every position, not just at the trajectory start. The aligner update (Eq. 12) runs after generation: future states are moved one small step against the gradient of ||Rφ(·) − y(τ)||^2, where Rφ is a trajectory-return regressor trained by Eq. 13; this pulls stochastic generations back toward the advertiser's stated objective without interacting with the denoising steps. An invers

Load-bearing premise

The approach assumes that the main reason earlier diffusion bidders failed is that their planned next states were unrealistic, and that training on randomly truncated histories fixes exactly that failure.

What would settle it

Run the AuctionNet-sparse experiment with the completer loss but replace the aligner's gradient update with a best-of-K rerank over sampled completions; if reranking matches CBD, the gradient form of the aligner is not the source of gain. Separately, measure dynamic legitimacy directly on generated states—via forward-model error or budget-monotonicity violations—for DiffBid, DiT-causal, and CBD-Completer; if legitimacy rates are similar while value differs, the completer is not doing the claimed causal work.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Diffusion-based auto-bidding can scale to large competitive auctions if planned states are constrained by observed history: the completer alone beats DiffBid and transformer or causal-attention modifications.
  • In sparse-reward auctions, a trajectory-level condition y(τ) supplies learning signal when per-step rewards are zero, yielding the reported 29.9% value improvement on AuctionNet-sparse at 100% budget.
  • The method transfers beyond auctions: D4RL locomotion results show consistent gains over diffuser baselines, suggesting masked-completion training is a general offline-RL fix.
  • On the Kuaishou MCB production, CBD raises target cost by 2.0% and CPA valid ratio by 1.85% at similar spend, with +6ms inference latency that fits within the 20s service budget.
  • The aligner enables controllable property alignment—such as smooth budget spend or avoiding early spend—which supports explanation and error correction in production bidding.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • An extension the paper leaves implicit: the same masked-completion loss could be applied to any diffusion planner, so a direct test is whether adding Lc to a causal-attention diffusion model closes the remaining gap with CBD.
  • Because the aligner refines only final outputs, it could be replaced by non-gradient post-processing; the paper's GS-Align reranking results nearly match gradient refinement, hinting that the trajectory-level objective matters more than the gradient update itself.
  • A sharper falsification of the causal story would measure dynamic legitimacy directly—for example, forward-model error or budget-monotonicity violations on generated states—across DiffBid, DiT-causal, and CBD-Completer, rather than relying on the paper's informally defined validity metric.
  • The 29.9% gain is computed against the baselines chosen; comparing CBD against a Decision Transformer with accumulated-return conditioning under the same inverse-dynamics bid mapping would isolate what the diffusion backbone itself contributes.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes CBD, a diffusion-based auto-bidding method for large-scale auctions, built from two components: a "completer" that trains the diffusion model with a masked future-only denoising loss (Eq. 11) while keeping a clean observed prefix s0:t, and an "aligner" that refines the generated future trajectory by gradient descent on a trajectory-level return model (Eq. 12). The authors claim that the completer addresses the "dynamic legitimacy" problem (generated states that are not reachable from the observed history) and that the aligner addresses misalignment with advertiser properties. Experiments on the AuctionNet and AuctionNet-sparse benchmarks report consistent improvements over RL, decision-transformer, and diffuser baselines, with the largest headline gain being a 29.9% improvement over DiffBid in the sparse setting at 100% budget. The paper also reports a 2.0% target-cost improvement in an online A/B test on Kuaishou and additional results on D4RL locomotion tasks.

Significance. If the central claims hold, the paper would make a useful contribution to generative auto-bidding: it identifies a concrete failure mode of diffuser-based planners in competitive auctions and proposes a simple, modular fix (masked completion training plus gradient-based refinement) that is compatible with arbitrary diffusion backbones and samplers. The use of a public benchmark (AuctionNet) and a large-scale online A/B test adds credibility. The two loss formulations are clearly stated and the ablations do separate the completer and aligner contributions from each other. However, the load-bearing claim that the completer improves performance specifically through dynamic legitimacy is not yet established, because the comparison confounds the masked loss with additional history conditioning, and the only quantitative validity metric is never defined. The headline improvement is also computed against the weakest diffuser baseline, while gains over the strongest prior methods are substantially smaller. These issues make the current evidence insufficient for the paper's stated causal narrative, though they are addressable with additional controlled experiments and clearer reporting.

major comments (5)
  1. [§4.2, Table 1] The abstract and Section 1 headline the 29.9% improvement, but this is measured against DiffBid, the weakest diffuser baseline. Relative to the strongest prior methods, the gains are much smaller: at AuctionNet-sparse 100% budget, CBD (40.71) improves over IQL (36.06) by about 12.9% and over DT-S (36.82) by about 10.6%; at AuctionNet 100%, CBD (374.0) is only 0.24% above DT-S (373.1). The paper should state the improvement over the best prior method, not only over DiffBid, and should avoid presenting the DiffBid-relative number as the representative gain.
  2. [§3.1, Eq. (11) and §4.4, Table 3] The central causal claim — that the completer's masked completion loss (Eq. 11) improves dynamic legitimacy — is not isolated from the change in input representation. CBD-Completer differs from DiffBid in two ways simultaneously: (i) the loss is computed only on positions t+1:T, and (ii) the model is conditioned on a clean observed prefix s0:t at every noise level (Eqs. 6–9), whereas DiffBid's inference places a single state at the first position and noisily generates the rest. No ablation holds the history-conditioning fixed while removing the masked loss. The only evidence for the legitimacy mechanism is Figure 2 and the 'Validity' column in Table 3, but the definition of Validity is never given, and it is not reported for CBD-Completer before the aligner is applied. Please add an ablation such as (a) DiffBid with a clean observed prefix and full-sequence loss, or (b) CBD with the loss
  3. [§3.2, Eq. (12) and Algorithm 2] There is a discrepancy between the aligner formula in Eq. (12) and its description in Algorithm 2, line 4. Eq. (12) uses ||R_φ(·) − y(τ)||², which moves the generated property toward the target y(τ). Algorithm 2 omits the target and writes ||R_φ(·)||², which would maximize the squared return norm rather than align to a specified property. This is a load-bearing inconsistency for reproducibility: as written, the algorithm does not implement the proposed aligner. Please correct the algorithm and specify whether the implementation follows Eq. (12).
  4. [§4.5, Table 4] The text states that 'the CBD method achieves state-of-the-art performance' on D4RL, but Table 4 does not support this uniformly. For example, Medium-Replay Walker2d: CBD 75.2 vs. DL 90.2; Medium Hopper: CBD 97.7 vs. DL 100.9; and several other cells are not the best. The claim should be either tempered to 'competitive with state-of-the-art on most tasks' or supported with a statistical aggregation (e.g., mean rank or paired tests across tasks).
  5. [§4.1 and §4.2] The paper reports that the improvements are 'statistically significant' with a two-sided t-test at p<0.05, but no standard deviations, confidence intervals, or number of runs are given for the AuctionNet results. Since the performance differences between CBD and DT-S are under 1% in several budget settings (e.g., AuctionNet 100%), a statistically meaningful comparison requires variance information. Please report the number of evaluation episodes, the standard errors, or confidence intervals, at least for the main comparisons.
minor comments (5)
  1. [§3.1, Eq. (11)] The distribution of the augmented variable t is not specified. The text says 't ∼ [0, T−1]' in Algorithm 1, but it is unclear whether this is uniform over all integers in the range. Please state the sampling distribution explicitly.
  2. [§4.4, Table 3] The 'Validity' metric is introduced without a definition. Even if a formal definition is added per Major Comment 2, please also state how it is computed (e.g., whether it checks monotonicity of remaining budget, or uses a learned forward model) and over how many generated trajectories.
  3. [§4.3, Figure 3] The property-alignment results are only visualized as distributions. Reporting a quantitative divergence measure (e.g., Wasserstein distance or a cdf-based statistic) between the aligned and non-aligned distributions would make the claim more precise and easier to verify.
  4. [Algorithm 2, line 4] The aligner step in Algorithm 2 also omits the target y(τ); this is part of Major Comment 3. Once corrected, please ensure the notation matches between Eq. (12), Algorithm 2, and the textual description in §3.2.
  5. [§4.5, Table 4] The D4RL results report mean and standard error over 150 episode seeds, but the standard errors are not shown for the non-diffuser baselines in the same table, making the comparison visually uneven. Consider reporting all values with the same precision.

Circularity Check

0 steps flagged

No significant circularity: the central results are empirical, and the only self-citation is motivational and not load-bearing.

full rationale

The paper's central claims are empirical evaluations on public AuctionNet benchmarks and a live Kuaishou A/B test. The two proposed components are standard conditional modeling operations: the completer loss in Eq. (11) is a masked denoising objective that trains the model to predict future states given a clean observed prefix, and the aligner in Eq. (12) is a post-hoc gradient refinement using a separately trained trajectory-level return model (Eq. (13)). Neither of these components reduces by construction to the reported evaluation metrics (Value, ER, Score, Target Cost), and the headline improvements are measured against external baselines and a real deployment, not derived from the model's own training objective. The only self-referential element is the citation of GAS [54] and GAVE [55] in Section 2.2 as evidence that DiffBid is ineffective in large-scale auctions. That claim is motivational and is independently supported by the paper's own comparisons of DiffBid against CBD in Tables 1 and 2, so the self-citation is not load-bearing. The undefined 'Validity' metric in Table 3 and the confound between added historical conditioning and the masked completion loss are experimental attribution/reproducibility concerns, not circularity: they do not show that a prediction is equivalent to a fitted input by definition. Overall, the derivation chain is self-contained and the empirical claims stand on their own measurements.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The method introduces no new physical or empirical entities; it reuses standard diffusion modeling, an inverse dynamics model, and a regression-based return model as components. The main assumptions are domain modeling choices (linear bid form, state features, inverse-model fidelity) and the two informal claims about dynamic legitimacy and return-model calibration, both of which are empirically motivated but not rigorously established.

free parameters (3)
  • Aligner gradient step size lambda = 0.1
    Chosen following prior diffuser settings and validated only on a portion of offline data (Section 4.4). A hand-chosen constant that directly controls the aligner update (Eq. 12).
  • Classifier-free guidance scalar omega = 1
    Set to 1 during inference 'to alleviate generation uncertainty' without an ablation (Section 3.1). Affects the diversity versus property-alignment trade-off.
  • Number of diffusion steps K = 100
    Selected via the ablation of Figure 4 as the smallest step count achieving good performance. A hyperparameter, not derived.
axioms (4)
  • domain assumption The optimal bid has the linear form b_i* = lambda0*v_i + sum_j lambda_j*p_ij*C_j (Eq. 2), and bidding is implemented by adjusting the lambda_j parameters periodically.
    Taken from USCB [23] as the standard industrial auto-bidding formulation (Section 2.1). All evaluated methods inherit this parameterized bid structure.
  • domain assumption A fixed-length trajectory of 48 states with the 16 features listed in Appendix B, plus an inverse dynamics model f_phi trained by Eq. (5), can map a generated next state to a valid bid adjustment.
    Core to the entire decision-diffuser paradigm (Eq. 5). If generated states leave the inverse-model training distribution, the actions derived from them degrade.
  • ad hoc to paper The masked future-loss objective of Eq. (11) is a valid surrogate for the conditional log-likelihood log p_theta(s_{t+1:T} | s0:t, y(tau)), and the resulting model generates dynamically legitimate states at inference.
    This is the central informal claim of the completer (Section 3.1), supported only by Figure 2 and an undefined validity metric. No proof is given that uniform random t training transfers to the inference-time prefix distribution.
  • ad hoc to paper The trajectory-level return model R_phi trained by Eq. (13) is accurate enough that its gradient at a generated trajectory points toward trajectories with better advertiser value.
    Used by the aligner in Eq. (12). No calibration or error analysis of R_phi is provided; the paper only reports downstream value improvements.

pith-pipeline@v1.4.0-alltime-deepseek-medium · 24528 in / 16552 out tokens · 155274 ms · 2026-08-05T10:58:28.650406+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Generative Auto-Bidding in Large-Scale Competitive Auctions via Diffusion Completer-Aligner." pith.science (2026). https://pith.science/paper/BKB4L5OW

@misc{pith2026250903348,
  author       = {Pith},
  title        = {Pith review of: Generative Auto-Bidding in Large-Scale Competitive Auctions via Diffusion Completer-Aligner},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BKB4L5OW}},
  note         = {Machine review of arXiv:2509.03348}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Auto-bidding is central to computational advertising, achieving notable commercial success by optimizing advertisers' bids within economic constraints. Recently, large generative models show potential to revolutionize auto-bidding by generating bids that could flexibly adapt to complex, competitive environments. Among them, diffusers stand out for their ability to address sparse-reward challenges by focusing on trajectory-level accumulated rewards, as well as their explainable capability, i.e., planning a future trajectory of states and executing bids accordingly. However, diffusers struggle with generation uncertainty, particularly regarding dynamic legitimacy between adjacent states, which can lead to poor bids and further cause significant loss of ad impression opportunities when competing with other advertisers in a highly competitive auction environment. To address it, we propose a Causal auto-Bidding method based on a Diffusion completer-aligner framework, termed CBD. Firstly, we augment the diffusion training process with an extra random variable t, where the model observes t-length historical sequences with the goal of completing the remaining sequence, thereby enhancing the generated sequences' dynamic legitimacy. Then, we employ a trajectory-level return model to refine the generated trajectories, aligning more closely with advertisers' objectives. Experimental results across diverse settings demonstrate that our approach not only achieves superior performance on large-scale auto-bidding benchmarks, such as a 29.9% improvement in conversion value in the challenging sparse-reward auction setting, but also delivers significant improvements on the Kuaishou online advertising platform, including a 2.0% increase in target cost.

Figures

Figures reproduced from arXiv: 2509.03348 by Bo An, Fei Pan, Jingtong Gao, Nan Jiang, Peng Jiang, Qingpeng Cai, Ruyi An, Shuai Mao, Xiangyu Zhao, Yewen Li.

Figure 1
Figure 1. Figure 1: Overview of our CBD method: At each timestep [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: visualization of the generation uncertainty issue using an example case. In all subfigures, the [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Results for property alignment. (a-b) show the generated trajectories with different [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Effect of the diffu￾sion steps. Our findings indicate that 100 denoising steps are sufficient for achieving good performance, resulting in a computationally efficient approach. Though industrial auto-bidding services have sufficient time for model’s response as stated in Section 2.1, we also introduce a distillation method as a post-training trick, named SiD [70], to distill our trained diffusion model int… view at source ↗
Figure 5
Figure 5. Figure 5: Real-time bidding system. To maximize the total value of impressions for an advertiser while adhering to certain economic constraints, research into advanced auto-bidding strategies has progressed from rule-based approaches to prediction-based methods, and more recently, to reinforcement learning-based techniques [76]. For rule-based methods, the proportional-integral-differential (PID) is the most widely … view at source ↗
Figure 6
Figure 6. Figure 6: Generated trajectories of DiffBid. 22 [PITH_FULL_IMAGE:figures/full_fig_p022_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Generated trajectories of CBD. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. D$^3$-Subsidy: Online and Sequential Driver Subsidy Decision-Making for Large-Scale Ride-Hailing Market

    cs.LG 2026-05 unverdicted novelty 7.0

    D³-Subsidy uses prefix-conditioned diffusion models to sample future trajectories and a Lagrangian mapping to produce cap-compliant city-level subsidy controls that improve rides and GMV in large ride-hailing systems.

  2. D$^3$-Subsidy: Online and Sequential Driver Subsidy Decision-Making for Large-Scale Ride-Hailing Market

    cs.LG 2026-05 unverdicted novelty 7.0

    D³-Subsidy uses prefix-conditioned diffusion to sample future trajectories from history, decodes them into city-level controls, and applies Lagrangian mapping to enforce subsidy caps, reporting gains in rides and GMV ...

  3. JD-BP: A Joint-Decision Generative Framework for Auto-Bidding and Pricing

    cs.GT 2026-04 unverdicted novelty 6.0

    JD-BP jointly generates bids and pricing corrections via generative models, memory-less return-to-go, trajectory augmentation, and energy-based DPO to improve auto-bidding performance despite prediction errors and latency.

  4. JD-BP: A Joint-Decision Generative Framework for Auto-Bidding and Pricing

    cs.GT 2026-04 reject novelty 6.0

    A generative auto-bidding model that jointly emits a bid and an additive price correction outperforms bid-only baselines on AuctionNet and online at JD.com.

  5. AIGB-R1: Self-Evolving Generative Auto-Bidding via Hierarchical Planner-Executor Optimization

    cs.LG 2026-07 conditional novelty 5.0

    AIGB-R1 couples an LLM strategy planner with a prompt-conditioned Decision Transformer executor and trains both end-to-end with a decoupled-GRPO self-evolving loop, reporting the highest Score on the AuctionNet benchmark.

  6. From Bootstrapping to Sequence Modeling: A Unified Generative Framework for Personalized Landing-Page Modeling

    cs.IR 2026-06 unverdicted novelty 5.0

    GLAN replaces CQL bootstrapping with Decision Transformer sequence modeling for PLPM, using global inter-day (L-RTG) and local session (HRM) modules to achieve +0.158% DAU and +0.108% LT gains in Kuaishou online tests.

  7. D$^3$-Subsidy: Online and Sequential Driver Subsidy Decision-Making for Large-Scale Ride-Hailing Market

    cs.LG 2026-05 unverdicted novelty 5.0

    D³-Subsidy uses prefix-conditioned diffusion models to generate future subsidy trajectories from history, decoded via inverse modules and mapped with Lagrangian duality to city-level incentives.

  8. D$^3$-Subsidy: Online and Sequential Driver Subsidy Decision-Making for Large-Scale Ride-Hailing Market

    cs.LG 2026-05 unverdicted novelty 5.0

    D³-Subsidy is a prefix-conditioned diffusion model plus Lagrangian mapping that generates city-level subsidy plans from historical data and maps them to per-order incentives while respecting rate caps.

Reference graph

Works this paper leans on

89 extracted references · 68 canonical work pages · cited by 4 Pith papers · 4 internal anchors

  1. [1]

    Real-time bidding: A new frontier of computational advertising research

    Jun Wang and Shuai Yuan. Real-time bidding: A new frontier of computational advertising research. In Proceedings of the Eighth ACM International Conference on Web Search and Data Mining, WSDM 2015, Shanghai, China, February 2-6, 2015, pages 415–416. ACM, 2015

    2015

  2. [2]

    The online advertising industry: Economics, evolution, and privacy

    David S Evans. The online advertising industry: Economics, evolution, and privacy. Journal of economic perspectives, 23(3):37–60, 2009

    2009

  3. [3]

    Muthukrishnan

    S. Muthukrishnan. Ad exchanges: Research issues. In Internet and Network Economics, 5th International Workshop, WINE 2009, Rome, Italy, December 14-18, 2009. Proceedings, volume 5929 of Lecture Notes in Computer Science, pages 1–12. Springer, 2009

    2009

  4. [4]

    Bidding machine: Learning to bid for directly optimizing profits in display advertising

    Kan Ren, Weinan Zhang, Ke Chang, Yifei Rong, Yong Yu, and Jun Wang. Bidding machine: Learning to bid for directly optimizing profits in display advertising. IEEE Transactions on Knowledge and Data Engineering, 30(4):645–659, 2017

    2017

  5. [5]

    Generative auto-bidding via conditional diffusion modeling

    Jiayan Guo, Yusen Huo, Zhilin Zhang, Tianyu Wang, Chuan Yu, Jian Xu, Bo Zheng, and Yan Zhang. Generative auto-bidding via conditional diffusion modeling. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, KDD 2024, Barcelona, Spain, August 25-29, 2024, pages 5038–5049. ACM, 2024

    2024

  6. [6]

    Optimized cost per click in taobao display advertising

    Han Zhu, Junqi Jin, Chang Tan, Fei Pan, Yifan Zeng, Han Li, and Kun Gai. Optimized cost per click in taobao display advertising. In Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining, pages 2191–2200, 2017

    2017

  7. [7]

    Aiads: Automated and intelligent advertising system for sponsored search

    Xiao Yang, Daren Sun, Ruiwei Zhu, Tao Deng, Zhi Guo, Zongyao Ding, Shouke Qin, and Yanfeng Zhu. Aiads: Automated and intelligent advertising system for sponsored search. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 1881–1890, 2019

    2019

  8. [8]

    Optimal bidding strategy for brand advertising

    Takanori Maehara, Atsuhiro Narita, Jun Baba, and Takayuki Kawabata. Optimal bidding strategy for brand advertising. In IJCAI, pages 424–432, 2018

    2018

  9. [9]

    Bid optimization by multivariable control in display advertising

    Xun Yang, Yasong Li, Hao Wang, Di Wu, Qing Tan, Jian Xu, and Kun Gai. Bid optimization by multivariable control in display advertising. In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, pages 1966–1974, 2019

    1966

  10. [10]

    Ad serving with multiple kpis

    Brendan Kitts, Michael Krishnan, Ishadutta Yadav, Yongbo Zeng, Garrett Badeau, Andrew Potter, Sergey Tolkachov, Ethan Thornburg, and Satyanarayana Reddy Janga. Ad serving with multiple kpis. In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1853–1861, 2017

    work page 2017

  11. [11]

    Joint optimization of multiple performance metrics in online video advertising

    Sahin Cem Geyik, Sergey Faleev, Jianqiang Shen, Sean O’Donnell, and Santanu Kolay. Joint optimization of multiple performance metrics in online video advertising. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 471–480, 2016

    work page 2016

  12. [12]

    Ye Chen, Pavel Berkhin, Bo Anderson, and Nikhil R. Devanur. Real-time bidding algorithms for performance-based display ad allocation. In Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Diego, CA, USA, August 21-24, 2011, pages 1307–1315. ACM, 2011

    work page 2011

  13. [13]

    Combining powers of two predictors in optimizing real-time bidding strategy under constrained budget

    Chi-Chun Lin, Kun-Ta Chuang, Wush Chi-Hsuan Wu, and Ming-Syan Chen. Combining powers of two predictors in optimizing real-time bidding strategy under constrained budget. In Proceedings of the 25th ACM International on Conference on Information and Knowledge Management, pages 2143–2148, 2016

    work page 2016

  14. [14]

    Meow: A space-efficient nonparametric bid shading algorithm

    Wei Zhang, Brendan Kitts, Yanjun Han, Zhengyuan Zhou, Tingyu Mao, Hao He, Shengjun Pan, Aaron Flores, San Gultekin, and Tsachy Weissman. Meow: A space-efficient nonparametric bid shading algorithm. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 3928–3936, 2021. 12

    work page 2021

  15. [15]

    An efficient deep distribution network for bid shading in first-price auctions

    Tian Zhou, Hao He, Shengjun Pan, Niklas Karlsson, Bharatbhushan Shetty, Brendan Kitts, Djordje Gligorijevic, San Gultekin, Tingyu Mao, Junwei Pan, et al. An efficient deep distribution network for bid shading in first-price auctions. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 3996–4004, 2021

    work page 2021

  16. [16]

    Bid shading in the brave new world of first-price auctions

    Djordje Gligorijevic, Tian Zhou, Bharatbhushan Shetty, Brendan Kitts, Shengjun Pan, Junwei Pan, and Aaron Flores. Bid shading in the brave new world of first-price auctions. InProceedings of the 29th ACM International Conference on Information & Knowledge Management, pages 2453–2460, 2020

    work page 2020

  17. [17]

    Lift-based bidding in ad selection

    Jian Xu, Xuhui Shao, Jianjie Ma, Kuang-chih Lee, Hang Qi, and Quan Lu. Lift-based bidding in ad selection. In Proceedings of the aaai conference on artificial intelligence , volume 30, 2016

    work page 2016

  18. [18]

    Attribution modeling increases efficiency of bidding in display advertising

    Eustache Diemert, Julien Meynet, Pierre Galland, and Damien Lefortier. Attribution modeling increases efficiency of bidding in display advertising. In Proceedings of the ADKDD’17, pages 1–6. 2017

    work page 2017

  19. [19]

    Causal models for real time bidding with repeated user interactions

    Martin Bompaire, Alexandre Gilotte, and Benjamin Heymann. Causal models for real time bidding with repeated user interactions. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 75–85, 2021

    work page 2021

  20. [20]

    Bid-aware gradient descent for unbiased learning with censored data in display advertising

    Weinan Zhang, Tianxiong Zhou, Jun Wang, and Jian Xu. Bid-aware gradient descent for unbiased learning with censored data in display advertising. In Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining, pages 665–674, 2016

    work page 2016

  21. [21]

    A cooperative-competitive multi-agent framework for auto-bidding in online advertising

    Chao Wen, Miao Xu, Zhilin Zhang, Zhenzhe Zheng, Yuhui Wang, Xiangyu Liu, Yu Rong, Dong Xie, Xiaoyang Tan, Chuan Yu, Jian Xu, Fan Wu, Guihai Chen, Xiaoqiang Zhu, and Bo Zheng. A cooperative-competitive multi-agent framework for auto-bidding in online advertising. In WSDM ’22: The Fifteenth ACM International Conference on Web Search and Data Mining, Virtual...

    work page 2022

  22. [22]

    Real-time bidding with multi-agent reinforcement learning in display advertising

    Junqi Jin, Chengru Song, Han Li, Kun Gai, Jun Wang, and Weinan Zhang. Real-time bidding with multi-agent reinforcement learning in display advertising. In Proceedings of the 27th ACM international conference on information and knowledge management, pages 2193–2201, 2018

    work page 2018

  23. [23]

    A unified solution to constrained bidding in online display advertising

    Yue He, Xiujun Chen, Di Wu, Junwei Pan, Qing Tan, Chuan Yu, Jian Xu, and Xiaoqiang Zhu. A unified solution to constrained bidding in online display advertising. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, pages 2993–3001, 2021

    work page 2021

  24. [24]

    Roi-constrained bidding via curriculum-guided bayesian reinforcement learning

    Haozhe Wang, Chao Du, Panyan Fang, Shuo Yuan, Xuming He, Liang Wang, and Bo Zheng. Roi-constrained bidding via curriculum-guided bayesian reinforcement learning. InProceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 4021– 4031, 2022

    work page 2022

  25. [25]

    Optimized cost per mille in feeds advertising

    Pingzhong Tang, Xun Wang, Zihe Wang, Yadong Xu, and Xiwang Yang. Optimized cost per mille in feeds advertising. In Proceedings of the 19th International Conference on Autonomous Agents and MultiAgent Systems, pages 1359–1367, 2020

    work page 2020

  26. [26]

    Rein- forcement learning with sequential information clustering in real-time bidding

    Junwei Lu, Chaoqi Yang, Xiaofeng Gao, Liubin Wang, Changcheng Li, and Guihai Chen. Rein- forcement learning with sequential information clustering in real-time bidding. In Proceedings of the 28th ACM International Conference on Information and Knowledge Management, pages 1633–1641, 2019

    work page 2019

  27. [27]

    Deep reinforcement learning for sponsored search real-time bidding

    Jun Zhao, Guang Qiu, Ziyu Guan, Wei Zhao, and Xiaofei He. Deep reinforcement learning for sponsored search real-time bidding. In Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining, pages 1021–1030, 2018

    work page 2018

  28. [28]

    Budget constrained bidding by model-free reinforcement learning in display advertising

    Di Wu, Xiujun Chen, Xun Yang, Hao Wang, Qing Tan, Xiaoxun Zhang, Jian Xu, and Kun Gai. Budget constrained bidding by model-free reinforcement learning in display advertising. In Proceedings of the 27th ACM International Conference on Information and Knowledge Management, pages 1443–1451, 2018. 13

    work page 2018

  29. [29]

    Real-time bidding by reinforcement learning in display advertising

    Han Cai, Kan Ren, Weinan Zhang, Kleanthis Malialis, Jun Wang, Yong Yu, and Defeng Guo. Real-time bidding by reinforcement learning in display advertising. In Proceedings of the tenth ACM international conference on web search and data mining, pages 661–670, 2017

    work page 2017

  30. [30]

    LADDER: A Human-Level Bidding Agent for Large-Scale Real-Time Online Auctions

    Yu Wang, Jiayi Liu, Yuxiang Liu, Jun Hao, Yang He, Jinghe Hu, Weipeng P Yan, and Mantian Li. Ladder: A human-level bidding agent for large-scale real-time online auctions. arXiv preprint arXiv:1708.05565, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  31. [31]

    Reinforcement learning: An introduction

    Richard S Sutton. Reinforcement learning: An introduction. A Bradford Book, 2018

    2018

  32. [32]

    OpenAI. Chatgpt. https://chatgpt.com/, 2024

    work page 2024

  33. [33]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. High-resolution image synthesis with latent diffusion models. In CVPR, 2022

    2022

  34. [34]

    Decision transformer: Reinforcement learning via sequence modeling

    Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, and Igor Mordatch. Decision transformer: Reinforcement learning via sequence modeling. In Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, December 6-14, 202...

    work page 2021

  35. [35]

    Constrained decision transformer for offline safe reinforcement learning

    Zuxin Liu, Zijian Guo, Yihang Yao, Zhepeng Cen, Wenhao Yu, Tingnan Zhang, and Ding Zhao. Constrained decision transformer for offline safe reinforcement learning. In International Conference on Machine Learning, ICML 2023, 23-29 July 2023, Honolulu, Hawaii, USA , volume 202 of Proceedings of Machine Learning Research, pages 21611–21630. PMLR, 2023

    work page 2023

  36. [36]

    Tenenbaum, Tommi S

    Anurag Ajay, Yilun Du, Abhi Gupta, Joshua B. Tenenbaum, Tommi S. Jaakkola, and Pulkit Agrawal. Is conditional generative modeling all you need for decision making? In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023, 2023

    work page 2023

  37. [38]

    Adapt- diffuser: Diffusion models as adaptive self-evolving planners

    Zhixuan Liang, Yao Mu, Mingyu Ding, Fei Ni, Masayoshi Tomizuka, and Ping Luo. Adapt- diffuser: Diffusion models as adaptive self-evolving planners. In International Conference on Machine Learning, ICML 2023, 23-29 July 2023, Honolulu, Hawaii, USA, volume 202 of Proceedings of Machine Learning Research, pages 20725–20745. PMLR, 2023

    work page 2023

  38. [39]

    Adaptive online replanning with diffusion models

    Siyuan Zhou, Yilun Du, Shun Zhang, Mengdi Xu, Yikang Shen, Wei Xiao, Dit-Yan Yeung, and Chuang Gan. Adaptive online replanning with diffusion models. In Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023, 2023

    work page 2023

  39. [40]

    Towards long-delayed sparsity: Learning a better transformer through reward redistribution

    Tianchen Zhu, Yue Qiu, Haoyi Zhou, and Jianxin Li. Towards long-delayed sparsity: Learning a better transformer through reward redistribution. In Proceedings of the Thirty-Second Interna- tional Joint Conference on Artificial Intelligence, IJCAI 2023, 19th-25th August 2023, Macao, SAR, China, pages 4693–4701. ijcai.org, 2023

    work page 2023

  40. [41]

    Auctionnet: A novel benchmark for decision-making in large-scale games

    Kefan Su, Yusen Huo, Zhilin Zhang, Shuai Dou, Chuan Yu, Jian Xu, Zongqing Lu, and Bo Zheng. Auctionnet: A novel benchmark for decision-making in large-scale games. In The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2024

    work page 2024

  41. [42]

    Cleandiffuser: An easy-to-use modularized library for diffusion models in decision making

    Zibin Dong, Yifu Yuan, Jianye Hao, Fei Ni, Yi Ma, Pengyi Li, and Yan Zheng. Cleandiffuser: An easy-to-use modularized library for diffusion models in decision making. In Advances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024, Vancouver, BC, Canada, December 10 - 15, 2024, 2024. 14

    work page 2024

  42. [43]

    Metadiffuser: Diffusion model as conditional planner for offline meta-rl

    Fei Ni, Jianye Hao, Yao Mu, Yifu Yuan, Yan Zheng, Bin Wang, and Zhixuan Liang. Metadiffuser: Diffusion model as conditional planner for offline meta-rl. In International Conference on Machine Learning, ICML 2023, 23-29 July 2023, Honolulu, Hawaii, USA , volume 202 of Proceedings of Machine Learning Research, pages 26087–26105. PMLR, 2023

    work page 2023

  43. [44]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. In Hugo Larochelle, Marc’Aurelio Ranzato, Raia Hadsell, Maria-Florina Balcan, and Hsuan-Tien Lin, editors, Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual, 2020

    work page 2020

  44. [45]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021, 2021

    work page 2021

  45. [46]

    D4RL: datasets for deep data-driven reinforcement learning

    Justin Fu, Aviral Kumar, Ofir Nachum, George Tucker, and Sergey Levine. D4RL: datasets for deep data-driven reinforcement learning. CoRR, abs/2004.07219, 2020

    Pith/arXiv arXiv 2004

  46. [47]

    Diffuserlite: Towards real-time diffusion planning

    Zibin Dong, Jianye Hao, Yifu Yuan, Fei Ni, Yitian Wang, Pengyi Li, and Yan Zheng. Diffuserlite: Towards real-time diffusion planning. InAdvances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024, Vancouver, BC, Canada, December 10 - 15, 2024, 2024

    work page 2024

  47. [48]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In IEEE/CVF International Conference on Computer Vision, ICCV 2023, Paris, France, October 1-6, 2023, pages 4172–4182. IEEE, 2023

    work page 2023

  48. [49]

    Causal attention for vision-language tasks

    Xu Yang, Hanwang Zhang, Guojun Qi, and Jianfei Cai. Causal attention for vision-language tasks. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021, pages 9847–9857. Computer Vision Foundation / IEEE, 2021

    work page 2021

  49. [50]

    Language models are unsupervised multitask learners

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. Language models are unsupervised multitask learners. OpenAI blog, 1(8):9, 2019

    2019

  50. [51]

    Aligner: Efficient alignment by learning to correct

    Jiaming Ji, Boyuan Chen, Hantao Lou, Donghai Hong, Borong Zhang, Xuehai Pan, Tianyi Qiu, Juntao Dai, and Yaodong Yang. Aligner: Efficient alignment by learning to correct. InAdvances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024, Vancouver, BC, Canada, December 10 - 15, 2024, 2024

    work page 2024

  51. [52]

    Attention is all you need

    A Waswani, N Shazeer, N Parmar, J Uszkoreit, L Jones, A Gomez, L Kaiser, and I Polosukhin. Attention is all you need. In NIPS, 2017

    2017

  52. [53]

    Sustainable online reinforcement learning for auto-bidding

    Zhiyu Mou, Yusen Huo, Rongquan Bai, Mingzhou Xie, Chuan Yu, Jian Xu, and Bo Zheng. Sustainable online reinforcement learning for auto-bidding. In Advances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing Systems 2022, NeurIPS 2022, New Orleans, LA, USA, November 28 - December 9, 2022, 2022

    work page 2022

  53. [54]

    GAS: Generative Auto-bidding with Post-training Search

    Yewen Li, Shuai Mao, Jingtong Gao, Nan Jiang, Yunjian Xu, Qingpeng Cai, Fei Pan, Peng Jiang, and Bo An. GAS: generative auto-bidding with post-training search. CoRR, abs/2412.17018, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  54. [55]

    Generative auto-bidding with value-guided explorations

    Jingtong Gao, Yewen Li, Shuai Mao, Peng Jiang, Nan Jiang, Yejing Wang, Qingpeng Cai, Fei Pan, Kun Gai, Bo An, et al. Generative auto-bidding with value-guided explorations. arXiv preprint arXiv:2504.14587, 2025

    work page arXiv 2025

  55. [56]

    Tenenbaum, and Sergey Levine

    Michael Janner, Yilun Du, Joshua B. Tenenbaum, and Sergey Levine. Planning with diffusion for flexible behavior synthesis. In International Conference on Machine Learning, ICML 2022, 17-23 July 2022, Baltimore, Maryland, USA, volume 162 of Proceedings of Machine Learning Research, pages 9902–9915. PMLR, 2022

    work page 2022

  56. [57]

    Diffusion forcing: Next-token prediction meets full-sequence diffusion

    Boyuan Chen, Diego Marti Monso, Yilun Du, Max Simchowitz, Russ Tedrake, and Vincent Sitzmann. Diffusion forcing: Next-token prediction meets full-sequence diffusion. In Advances in Neural Information Processing Systems 38: Annual Conference on Neural Information Processing Systems 2024, NeurIPS 2024, Vancouver, BC, Canada, December 10 - 15, 2024, 2024. 15

    work page 2024

  57. [58]

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025

    Pith/arXiv arXiv 2025

  58. [59]

    Qwen technical report

    Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, Binyuan Hui, Luo Ji, Mei Li, Junyang Lin, Runji Lin, Dayiheng Liu, Gao Liu, Chengqiang Lu, Keming Lu, Jianxin Ma, Rui Men, Xingzhang Ren, Xuancheng Ren, Chuanqi Tan, Sinan Tan, Jianhong Tu, Peng Wang, Shijie Wang, Wei Wang, Shengguang Wu, Benfeng X...

    Pith/arXiv arXiv 2023

  59. [60]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. CoRR, abs/2207.12598, 2022

    Pith/arXiv arXiv 2022

  60. [61]

    Aligndiff: Aligning diverse human preferences via behavior- customisable diffusion model

    Zibin Dong, Yifu Yuan, Jianye Hao, Fei Ni, Yao Mu, Yan Zheng, Yujing Hu, Tangjie Lv, Changjie Fan, and Zhipeng Hu. Aligndiff: Aligning diverse human preferences via behavior- customisable diffusion model. In The Twelfth International Conference on Learning Represen- tations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net, 2024

    work page 2024

  61. [62]

    Improved techniques for maximum likelihood estimation for diffusion odes

    Kaiwen Zheng, Cheng Lu, Jianfei Chen, and Jun Zhu. Improved techniques for maximum likelihood estimation for diffusion odes. In International Conference on Machine Learning, ICML 2023, 23-29 July 2023, Honolulu, Hawaii, USA, volume 202 of Proceedings of Machine Learning Research, pages 42363–42389. PMLR, 2023

    work page 2023

  62. [63]

    Dpm-solver-v3: Improved diffusion ODE solver with empirical model statistics

    Kaiwen Zheng, Cheng Lu, Jianfei Chen, and Jun Zhu. Dpm-solver-v3: Improved diffusion ODE solver with empirical model statistics. In Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New Orleans, LA, USA, December 10 - 16, 2023, 2023

    work page 2023

  63. [64]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. OpenReview.net, 2022

    work page 2022

  64. [65]

    Off-policy deep reinforcement learning without exploration

    Scott Fujimoto, David Meger, and Doina Precup. Off-policy deep reinforcement learning without exploration. In Proceedings of the 36th International Conference on Machine Learning, ICML 2019, 9-15 June 2019, Long Beach, California, USA , volume 97 of Proceedings of Machine Learning Research, pages 2052–2062. PMLR, 2019

    work page 2019

  65. [66]

    Conservative q-learning for offline reinforcement learning

    Aviral Kumar, Aurick Zhou, George Tucker, and Sergey Levine. Conservative q-learning for offline reinforcement learning. In Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual, 2020

    work page 2020

  66. [67]

    Offline reinforcement learning with implicit q-learning

    Ilya Kostrikov, Ashvin Nair, and Sergey Levine. Offline reinforcement learning with implicit q-learning. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022, 2022

    2022

  67. [68]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Inter- vention - MICCAI 2015 - 18th International Conference Munich, Germany, October 5 - 9, 2015, Proceedings, Part III, volume 9351 of Lecture Notes in Computer Science , pages 234–241. Spr...

    work page 2015

  68. [69]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In International Conference on Learning Representations

    work page

  69. [70]

    Score identity distillation: Exponentially fast distillation of pretrained diffusion models for one-step generation

    Mingyuan Zhou, Huangjie Zheng, Zhendong Wang, Mingzhang Yin, and Hai Huang. Score identity distillation: Exponentially fast distillation of pretrained diffusion models for one-step generation. In Forty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024, 2024. 16

    work page 2024

  70. [71]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V . Le, and Denny Zhou. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing Systems 2022, NeurIPS 2022, New Orleans, LA, USA, November ...

    work page 2022

  71. [72]

    Iab internet advertising revenue report 2024

    International Advertising Bureau. Iab internet advertising revenue report 2024. https://www. iab.com/, 2024

    work page 2024

  72. [73]

    Dave and Vasudeva Varma

    Kushal S. Dave and Vasudeva Varma. Computational advertising: Techniques for targeting relevant ads. Found. Trends Inf. Retr., 8(4-5):263–418, 2014

    work page 2014

  73. [74]

    Display advertising with real-time bidding (RTB) and behavioural targeting

    Jun Wang, Weinan Zhang, and Shuai Yuan. Display advertising with real-time bidding (RTB) and behavioural targeting. Found. Trends Inf. Retr., 11(4-5):297–435, 2017

    work page 2017

  74. [75]

    Kearns, Peter B

    Kareem Amin, Michael J. Kearns, Peter B. Key, and Anton Schwaighofer. Budget optimization for sponsored search: Censored learning in mdps. In Proceedings of the Twenty-Eighth Conference on Uncertainty in Artificial Intelligence, Catalina Island, CA, USA, August 14-18, 2012, pages 54–63. AUAI Press, 2012

    work page 2012

  75. [76]

    A survey on bid optimization in real-time bidding display advertising

    Weitong Ou, Bo Chen, Xinyi Dai, Weinan Zhang, Weiwen Liu, Ruiming Tang, and Yong Yu. A survey on bid optimization in real-time bidding display advertising. ACM Trans. Knowl. Discov. Data, 18(3):58:1–58:31, 2024

    work page 2024

  76. [77]

    Development of the pid controller

    Stuart Bennett. Development of the pid controller. IEEE Control Systems Magazine, 13(6): 58–62, 1993

    work page 1993

  77. [78]

    Recurrent neural networks for stochastic control in real-time bidding

    Nicolas Grislain, Nicolas Perrin, and Antoine Thabault. Recurrent neural networks for stochastic control in real-time bidding. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2019, Anchorage, AK, USA, August 4-8, 2019, pages 2801–2809. ACM, 2019

    work page 2019

  78. [79]

    Roger B. Myerson. Optimal auction design. Math. Oper. Res., 6(1):58–73, 1981

    work page 1981

  79. [80]

    User response learning for directly optimizing campaign performance in display advertising

    Kan Ren, Weinan Zhang, Yifei Rong, Haifeng Zhang, Yong Yu, and Jun Wang. User response learning for directly optimizing campaign performance in display advertising. In Proceedings of the 25th ACM International Conference on Information and Knowledge Management, CIKM 2016, Indianapolis, IN, USA, October 24-28, 2016, pages 679–688. ACM, 2016

    work page 2016

  80. [81]

    Deep neural net with attention for multi-channel multi-touch attribution

    Ning li, Sai Kumar Arava, Chen Dong, Zhenyu Yan, and Abhishek Pani. Deep neural net with attention for multi-channel multi-touch attribution. CoRR, abs/1809.02230, 2018. URL http://arxiv.org/abs/1809.02230

    Pith/arXiv arXiv 2018

Showing first 80 references.