Tagged articles

9 articles

Page 1 of 1

Oct 24, 2024 · Artificial Intelligence

Pre‑Ranking in Recommendation Systems: Model and Sample Optimization Practices at Zhuanzhuan Home Page

This article reviews the role of pre‑ranking in multi‑stage recommendation pipelines, compares dual‑tower and fully‑connected DNN models, discusses negative and positive sample selection strategies, and presents Zhuanzhuan's practical improvements in model architecture and traffic‑pool allocation to boost precision and diversity.

Model Optimizationdual-towerpre‑ranking

0 likes · 16 min read

Pre‑Ranking in Recommendation Systems: Model and Sample Optimization Practices at Zhuanzhuan Home Page

DataFunTalk

Feb 17, 2023 · Artificial Intelligence

Full‑Chain Linkage Techniques for Alibaba Mama Display Advertising: From Precise Value Estimation to Set‑Selection Models

The article presents a comprehensive technical roadmap for Alibaba Mama's display advertising cascade ranking system, introducing full‑chain linkage, precise‑value estimation models (PDM, ESDM) and set‑selection approaches (LDM, LBDM), and demonstrates how these innovations jointly improve CTR and RPM while outlining future research directions.

Advertisingmachine learningpre‑ranking

0 likes · 25 min read

Full‑Chain Linkage Techniques for Alibaba Mama Display Advertising: From Precise Value Estimation to Set‑Selection Models

Meituan Technology Team

Aug 18, 2022 · Artificial Intelligence

Highlights of Meituan’s KDD 2022 Papers

This article presents concise introductions and download links for seven Meituan research papers accepted at KDD 2022, covering knowledge‑graph pre‑training, automatic feature and architecture selection for pre‑ranking, user intent discovery, persuasion factor modeling, counterfactual policy learning for top‑K recommendation, probabilistic forecasting of food preparation time, and a multi‑stage bonus allocation framework for meal delivery.

KDD 2022Knowledge Graphbonus allocation

0 likes · 15 min read

DataFunSummit

Aug 14, 2022 · Artificial Intelligence

Optimizing Pre‑Ranking in Meituan Search: Knowledge Distillation and Neural Architecture Search

This article describes Meituan Search's pre‑ranking (coarse‑ranking) system evolution and presents two major optimization strategies—leveraging knowledge distillation to align coarse‑ranking with fine‑ranking and employing neural architecture search to jointly improve effectiveness and latency—demonstrating significant offline and online performance gains.

Neural Architecture Searchknowledge distillationmachine learning

0 likes · 17 min read

Optimizing Pre‑Ranking in Meituan Search: Knowledge Distillation and Neural Architecture Search

Meituan Technology Team

Aug 11, 2022 · Artificial Intelligence

Optimizing Pre‑Ranking in Meituan Search: Knowledge Distillation and Neural Architecture Search

Meituan’s search team upgraded its pre‑ranking layer from simple linear models to end‑to‑end neural networks, boosting effectiveness by applying three knowledge‑distillation techniques—including result‑list, score, and contrastive representation transfer—and by using latency‑aware neural architecture search to automatically select features and network structures, achieving significant recall and CTR gains without added latency.

Neural Architecture Searchefficiency optimizationknowledge distillation

0 likes · 19 min read

DataFunSummit

Sep 10, 2021 · Artificial Intelligence

Advances in Pre‑Ranking: The COLD System for Large‑Scale Advertising

This article reviews the evolution of coarse‑ranking in large‑scale ad systems, explains the two main technical routes—set selection and precise value estimation—introduces the Computing‑Power‑Cost‑Aware Online Lightweight Deep (COLD) pre‑ranking framework, and presents experimental results and future directions for deeper integration with fine‑ranking.

AdvertisingCOLDfeature selection

0 likes · 21 min read

Advances in Pre‑Ranking: The COLD System for Large‑Scale Advertising

Alimama Tech

May 27, 2021 · Artificial Intelligence

Towards a Better Tradeoff between Effectiveness and Efficiency in Pre‑Ranking: A Learnable Feature‑Selection‑Based Approach

The authors introduce an interaction‑focused pre‑ranking model combined with a learnable, complexity‑aware feature‑selection technique (FSCD) that selects a compact feature set, enabling Alibaba’s search advertising system to boost offline AUC from 0.695 to 0.737, raise recall to 95 %, improve CTR and RPM, yet retain CPU usage and latency comparable to traditional vector‑dot models.

effectivenessfeature selectionpre‑ranking

0 likes · 15 min read

Towards a Better Tradeoff between Effectiveness and Efficiency in Pre‑Ranking: A Learnable Feature‑Selection‑Based Approach

DataFunTalk

Feb 21, 2021 · Artificial Intelligence

Advances in Pre‑Ranking for Large‑Scale Advertising: The COLD Framework and Its Technical Evolution

This article reviews the development history, technical routes, and recent breakthroughs of pre‑ranking (coarse ranking) in large‑scale advertising systems, focusing on Alibaba's COLD (Computing‑power‑cost‑aware Online and Lightweight Deep) framework, its model design, engineering optimizations, experimental results, and future research directions.

AdvertisingCOLDOnline Learning

0 likes · 20 min read

Advances in Pre‑Ranking for Large‑Scale Advertising: The COLD Framework and Its Technical Evolution

DataFunTalk

Aug 18, 2020 · Artificial Intelligence

COLD: A Next‑Generation Pre‑Ranking System for Online Advertising

The article introduces COLD, a computing‑power‑aware online and lightweight deep pre‑ranking system for Alibaba's targeted ads, detailing its evolution from static CTR models to vector‑inner‑product models, its flexible network architecture with feature‑selection via SE blocks, engineering optimizations such as parallelism, column‑wise computation, Float16 and MPS, and demonstrates superior offline and online performance through extensive experiments.

COLDModel Optimizationfeature selection

0 likes · 11 min read

COLD: A Next‑Generation Pre‑Ranking System for Online Advertising