How UNIFIEDREWARD Breaks Task Boundaries to Boost Image and Video Performance

The paper introduces UNIFIEDREWARD, the first unified reward model for multimodal understanding and generation that supports pairwise ranking and pointwise scoring, builds a 236K human‑preference dataset across image and video tasks, and uses DPO to align VLMs and diffusion models, achieving significant performance gains on both image and video benchmarks.

Direct Preference OptimizationMultimodal AIPreference Modeling

0 likes · 19 min read

How UNIFIEDREWARD Breaks Task Boundaries to Boost Image and Video Performance

JD Tech

Aug 26, 2024 · Artificial Intelligence

Preference-oriented Diversity Model Based on Mutual Information for E-commerce Search Re-ranking (SIGIR 2024)

This article presents the SIGIR 2024 accepted PODM‑MI model, which uses variational inference and mutual‑information maximization to jointly optimize relevance and diversity in JD e‑commerce search re‑ranking, demonstrating significant gains in both user conversion and result diversity through extensive online experiments.

DiversityE-commerce SearchPreference Modeling

0 likes · 11 min read

Preference-oriented Diversity Model Based on Mutual Information for E-commerce Search Re-ranking (SIGIR 2024)

JD Tech Talk

Aug 26, 2024 · Artificial Intelligence

Preference-oriented Diversity Model Based on Mutual Information for E-commerce Re-ranking (SIGIR 2024)

The paper proposes PODM‑MI, a mutual‑information‑driven, preference‑oriented diversity model that jointly optimizes accuracy and diversity in e‑commerce search re‑ranking by modeling user preferences with multivariate Gaussian distributions and adapting rankings via a learnable utility matrix, showing significant gains in JD's main search experiments.

AIDiversityE-commerce Search

0 likes · 12 min read

Preference-oriented Diversity Model Based on Mutual Information for E-commerce Re-ranking (SIGIR 2024)

JD Cloud Developers

Aug 26, 2024 · Artificial Intelligence

Boosting E‑Commerce Re‑ranking Diversity and Accuracy with Mutual‑Information

This paper introduces PODM‑MI, a preference‑oriented diversity model that jointly optimizes relevance and diversity in e‑commerce search re‑ranking by leveraging variational inference and mutual‑information, demonstrating significant gains in both user conversion and result variety on JD.com.

DiversityE-commerce SearchPreference Modeling

0 likes · 10 min read

Boosting E‑Commerce Re‑ranking Diversity and Accuracy with Mutual‑Information

JD Retail Technology

Aug 26, 2024 · Artificial Intelligence

Preference-oriented Diversity Model Based on Mutual Information for E-commerce Search Re-ranking (SIGIR 2024)

This article introduces PODM‑MI, a preference‑oriented diversity model that uses mutual information and variational Gaussian representations to jointly optimize accuracy and diversity in e‑commerce search re‑ranking, and reports significant online A/B test improvements on JD.com.

DiversityPreference Modelinge‑commerce

0 likes · 10 min read