Papers for

advertising platform developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Ai judges help humans review and edit product video ads efficiently

Multimodal Judges Flag, Humans Decide: A Review-and-Edit Framework for AI-Generated Product Video Ads

Abstract: Generative pipelines can now turn every product page of an e-commerce catalog into a short video ad, but deciding whether such an ad is usable is not one good-or-bad call. We describe how we operationalized the review and editing of these ads so that both can keep pace with generation: reviewers go through each ad scene by scene and answer six questions, on the product, scene quality, pronunciation, script, sound and captions; judges built from vision-language, speech and language models file each finding under the matching question, and reviewers accept or reject each finding and add what the judges missed. Each confirmed issue then opens an edit tool matched to it. The questions and the routing from a finding to its fix do not depend on the models that generate the ad. Across 120 ads from our pipeline, reviewed with judge findings on screen, reviewers marked the product wrong on 60% of scenes, and script, caption or sound problems on under 10% each. On 60 of these ads, they marked the product wrong on 58% of scenes, far more than the 26% marked by two reviewers who later saw the ads with the judges hidden; one reviewer who saw the ads both ways marked 57% with judge findings and 19% without. On 120 blind-labeled scenes, a product judge that watches the clip with the catalog photos reached AUC 0.84, while video-quality and photo-similarity scores were close to chance. We tune the judges for recall: the product judge catches nearly all defects but also flags many clean scenes, so reviewers decide; together, the judges widen what reviewers mark and reviewers add what the judges miss. Without judge findings, one reviewer's marking faded over a session. We release a review-portal template and our review data (120 ads and 562 scenes with judge findings and human marks, plus 120 blind-labeled scenes with baseline scores) at https://github.com/anoubhav/ad-review-and-edit.

Mon 9 NovComputer Vision and Pattern RecognitionHuman-Computer InteractionMultimedia
The gist
Creating short video ads automatically for products is easy, but deciding if the videos are good enough is tricky. The authors built a system where AI judges first find possible problems with video ads, like the product shown being wrong or the script having issues. Then, human reviewers check those flagged problems, agree or disagree, and fix them using special editing tools. This cooperation helps catch more errors and keep up with how quickly videos are made.
Open → 2611.13178v1

Llm persona simulation improves cold start ad performance predictions

What Matters for LLM Persona Simulation: Signal Form in Cold-Start E-Commerce Advertising

Abstract: Estimating advertising performance in cold-start settings is challenging because historical logs are unavailable before launch. Large language models (LLMs) offer a potential alternative by simulating persona-conditioned user responses, but it remains unclear which design choices actually drive LLM-based persona simulation. Using real Google Ads e-commerce campaign data across five demographic segments, we identify three signal-form dimensions that affect simulation quality: (i) a demographic anchor without noisy rich attributes, (ii) a relative qualitative banner-blindness signal rather than quantitative anchors, and (iii) prompt phrasing. On this campaign, our method attains an MAE of 24.9 and a MAPE of 26.1%, a 51.3% lower MAE than the best tested non-LLM baseline and 77.3% lower than an LLM configuration using explicit CTR anchors. Systematic ablations show that minimal demographic descriptions achieve lower mean error than richer persona configurations, while relative qualitative cues substantially outperform the tested quantitative anchors. We additionally compare these configurations with direct-prior and rule-based baselines and with alternative LLM configurations.

Fri 6 NovMachine LearningArtificial Intelligence
The gist
Predicting how well ads will do before they launch is hard without past data. The authors studied how large language models can simulate user responses based on simple persona descriptions. They found that using minimal demographic info without noisy details, along with relative rather than exact cues, and specific prompt wording, makes predictions more accurate. Their approach significantly outperforms other non-LLM and LLM methods at estimating ad click-through rates in a real e-commerce setting.
Open → 2611.08840v1

Fair ranking methods ensure equal exposure for user groups and categories

Hidden Disparities in Ranking: Category-Aware Fairness via Distributional Stability

Abstract: We study the problem of fair ranking in recommender systems. We provide a formal probabilistic framework in which ranking policies are distributions over permutations, and the utility of a user is the exposure given to the items they interact with, relative to that of an ideal ranking. We consider utility parity across user groups, as well as its category-aware refinement, conditional utility parity, which requires parity within each category of items. Following a recent view of fairness as distributional stability, we show that both criteria are equivalent to the stability of the utility under changes in the composition of user groups. Optimizing over mixtures of ranking policies leads to a second-order cone program, for which we provide generalization bounds. A numerical study validates our theoretical findings.

Thu 29 OctMachine Learning
The gist
Recommender systems often show different sets of items to different users, which can lead to unfair exposure for certain groups. This study defines fairness as giving equal attention to all users and within each category of items users interact with. The researchers found fairness can be understood as keeping user satisfaction stable even when the mix of user groups changes. They created a mathematical method to find fair ranking strategies and tested it to confirm their theory.
Open → 2610.40182v1

Persistent candidate sets speed up linear bandit decision making

Accelerating Linear Bandits with Persistent Candidate Sets

Abstract: Exhaustive optimistic scoring can make action selection costly in linear bandits with a large fixed set of $K$ actions, requiring $\mathcal{O}(K)$ score evaluations per round. We introduce Persistent Candidate LinUCB (\textsc{PC-LinUCB}), which maintains $k$ active actions and samples $q$ challengers uniformly from the remaining $K-k$ actions at each round. By scoring only these $k+q$ candidates, \textsc{PC-LinUCB} reduces the number of score evaluations per round to $\mathcal{O}(k+q)$. We establish problem-independent and problem-dependent high-probability regret bounds consisting of a standard linear-bandit learning term and an additive discovery cost of order $\mathcal{O}\left((K-k)\log(K-k)/q\right)$ at fixed confidence. Notably, this additional cost does not grow with the horizon. For fixed gaps and sufficiently long horizons, we establish an $Ω\left((K-k)/q\right)$ regret lower bound under uniform challenger sampling, showing that partial candidate access itself imposes an unavoidable regret cost. Experiments show that \textsc{PC-LinUCB} substantially accelerates LinUCB while maintaining comparable regret, and that the persistent candidate mechanism can extend beyond LinUCB.

Fri 9 OctMachine Learning
The gist
Choosing the best option among many possibilities can be slow when each choice needs to be scored every time. The authors of this paper created a method called Persistent Candidate LinUCB that keeps a smaller set of active options and only checks a few new ones each round, making the process faster. They show that this approach still learns well with only a small extra cost that doesn’t grow over time. Their experiments confirm it speeds up decision making without losing quality.
Open → 2610.13407v1

Ecommerce ad images enhanced by targeted click-driven editing

Evolving Creatives with Clicks: CTR-Driven Advertising image Editing via Attribute Disentanglement and Preference Alignment

Abstract: The optimization objective for e-commerce advertising creatives has shifted from aesthetic quality to online business metrics like click-through rate (CTR). Recent methods generate appealing ad images from product images by constructing CTR-based reward signals and refining prompt generation via Reinforcement Learning. Such paradigms equally penalize all attributes within an underperforming image, even though some are actually beneficial to clicks. In this work, we propose Evolving Creatives with Clicks (ECC), which reformulates creative optimization as fine-grained editing rather than generation from scratch. ECC selectively refines attributes that hurt CTR while preserving those that contribute positively. Specifically, faced with a rich attribute space across massive images, ECC first identifies 28 attributes significantly correlated with CTR via ANOVA and Random Forest regression. Based on attribute contrasts between images with disparate CTRs, we instruction-tune a prompt model to generate editing suggestions, then refine it with real click preferences since reference-derived suggestions are not uniformly beneficial. Furthermore, we introduce Consistent Preference Optimization (CPO) as ECC's preference optimization strategy, which enforces consistency across semantically equivalent shuffled suggestions to mitigate DPO overfitting and improve preference generalization. Extensive experiments demonstrate that ECC significantly improves the CTR of ad images, outperforming state-of-the-art models by a substantial margin. The complete data pipeline, including attribute extraction and analysis, the resulting dataset, and source code will be released upon acceptance.

Fri 9 OctComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Online advertising aims to increase how often people click on ads instead of just making the images look nice. The authors found that changing every part of an underperforming ad image is not ideal since some parts actually help clicks. They created a method called ECC that focuses only on tweaking the parts of the image that reduce clicks while keeping the good ones. This approach uses statistical analysis to find key image traits linked to clicks and learns from real click patterns to improve advertising images. Their tests show ECC makes ad images get clicked on much more than other methods.
Open → 2610.13066v1

Embedding subspaces enable flexible multi-goal recommendations at scale

Embedding Subspace Partitioning for Dynamic Multi-Objective Retrieval

Abstract: Modern industrial recommender systems must optimize across competing objectives, balancing semantic relevance with business metrics such as engagement and revenue. While bi-encoders dominate large-scale retrieval due to their efficiency, they collapse these heterogeneous signals into a single static embedding space. This design creates a fundamental limitation: once trained, the retriever cannot adapt to shifting objective priorities at serving time without retraining. Moreover, joint optimization with multi-objective losses often induces interference between objectives, leading to suboptimal trade-offs. We propose Embedding Subspace Partitioning (ESP), a retrieval framework that decomposes the embedding into task-aware subspaces and replaces the single dot product with a weighted sum of per-subspace similarities, whose weights are tunable at serving time. For Transformer bi-encoders, ESP uses the model's native end-of-sequence token as a segment delimiter, with segment-aware attention masking and position encoding resets to guarantee subspace isolation in a single forward pass. Serving is performed via GPU-accelerated exhaustive kNN over one concatenated index, eliminating the need for per-objective Approximate Nearest Neighbor (ANN) infrastructure required by multi-head approaches. We evaluate ESP on an open-source benchmark built from MS MARCO. A single ESP model traces a broad Pareto frontier, consistently outperforming strong multi-task baselines across diverse operating points. In LinkedIn's job matching platform (70M+ weekly users), ESP enabled dynamic retrieval reconfiguration and delivered significant key business metric lifts.

Thu 24 SeptInformation Retrieval
The gist
Recommender systems often need to balance different goals like showing relevant items and increasing business value, but traditional methods mix all goals into one fixed understanding. The authors propose splitting the understanding into separate parts, each focused on a specific goal, so the system can adjust priorities on the fly without retraining. Their method works efficiently on large-scale systems by doing one search over combined data rather than multiple searches. Tests show this approach better balances goals and improved user engagement in a large job matching platform.
Open → 2609.30601v1

Parameter inheritance improves scaling and efficiency of recommendation models

Inherit4Rec: Parameter Inheritance for Efficient Scaling of Recommendation Models

Abstract: Scaling model capacity has emerged as an effective approach to overcoming performance bottlenecks in industrial recommender systems. However, repeatedly training larger dense models from scratch demands substantial data and time, while their growing computation conflicts with the strict serving budgets of industrial systems. Parameter inheritance provides a promising route for both dense model growth and sparse conversion, yet existing methods are primarily designed for static corpora and can suffer sharp performance drops under dynamically evolving recommendation data. To address these challenges, we propose Inherit4Rec, a parameter-inheritance framework that supports both Dense-to-Dense (D2D) growth and Dense-to-Sparse (D2S) conversion. Inherit4Rec-D2D combines hybrid growth with asymmetric training to preserve the forward function at expansion and maintain update continuity. Inherit4Rec-D2S constructs SMoE networks through co-activation-aware partitioning and a load-balancing loss, preserving dense-model capabilities while promoting balanced expert activation. Experiments on KuaiRand-1K and an industrial short-video recommendation dataset show that both transformations consistently outperform the evaluated inheritance baselines across all prediction objectives. These results demonstrate the effectiveness of Inherit4Rec for continual capacity expansion and computation-efficient sparse conversion in industrial recommender systems.

Sat 19 SeptInformation Retrieval
The gist
Recommender systems often need bigger models to improve their suggestions but training large models from scratch takes a lot of time and computing power. The authors propose a method called Inherit4Rec that helps grow models or make them sparser by reusing parameters from smaller models, so training is faster and requires less data. Their approach also handles changing recommendation data better than previous methods. Tests on public and industrial datasets show Inherit4Rec works well to keep improving recommendations while saving computation.
Open → 2609.23111v1

Fractional assignment methods ensure fair and strategic object distributions

Fractional Assignment with $\ell_1$ Preferences

Abstract: We study a fractional assignment setting where $n$ objects are to be assigned to $n$ agents with unit capacity, and each agent specifies an ideal distribution over the objects. Unlike in classic random assignment, these ideal distributions are not necessarily degenerate, as agents may prefer a mixture of objects rather than any single object. We assume that agents seek to minimize the $\ell_1$ distance between their ideal distribution and the distribution they receive, which is equivalent to maximizing the overlap between the two distributions. We propose two mechanisms, one based on water filling (WF) and the other on quadratic programming (QP), and show that both mechanisms are utilitarian-optimal (and hence Pareto efficient), envy-free, strategyproof, and satisfy equal treatment of equals. Moreover, we highlight a distinct advantage of each mechanism: while the WF mechanism satisfies the stronger property of group-strategyproofness, the QP mechanism is more robust in terms of egalitarian overlap welfare.

Wed 16 SeptComputer Science and Game Theory
The gist
This paper looks at a situation where multiple people want to share items in parts rather than getting just one whole item. Instead of each person wanting only one item, they may want a mix, and the goal is to match these preferences as closely as possible. The authors provide two ways to do this so that everyone feels treated fairly, no one envies others’ shares, and people cannot benefit by lying about their preferences. One method is better at preventing groups from cheating, while the other improves overall fairness among individuals.
Open → 2609.18299v1