Official Alimama tech channel, showcasing all of Alimama's technical innovations.
AIGB, introduced by Alibaba Mama in 2023, reframes large‑scale ad‑auction auto‑bidding as a generative sequence task using diffusion models, achieving up to 5 % GMV gains, improved stability and interpretability, and is now commercialized, open‑sourced, and featured in a NeurIPS‑endorsed competition.
FlowDCN introduces Groupwise‑MSDCN, a sparse deformable convolution that replaces attention, enabling efficient arbitrary‑resolution image generation with linear complexity, fewer parameters and FLOPs, and achieving state‑of‑the‑art FID scores on ImageNet while requiring far fewer training steps.
Alibaba's anti-fraud platform DeepString uses large unsupervised models to detect abnormal traffic in real time across multiple advertising products, combining a foundation model for event mining, anomaly measurement, and an alignment model for online filtering, reducing reliance on manual labeling and domain expertise.
This article presents the research behind Synchronized Video Storytelling, introducing the E‑SyncVidStory dataset, the VideoNarrator multimodal architecture, and extensive experiments that demonstrate high‑quality, product‑aware video narration generation for e‑commerce applications.
Alibaba’s Mama Intelligent Creation team has open‑sourced a FLUX‑based ControlNet inpainting model that leverages a DiT‑backed Interleave design for superior repair quality, and an 8‑step LoRA‑Turbo model that cuts inference time three‑fold while preserving near‑original image fidelity, both now available on Hugging Face and ModelScope.
The paper introduces a coupled generative adversarial framework that merges biased observational with unbiased experimental data to create a bias‑free dataset for causal inference, enabling robust treatment‑effect estimation under collider bias from an out‑of‑distribution perspective, and demonstrates superior bias reduction on three public advertising datasets.
This article presents a bi‑objective inventory allocation model for guaranteed‑delivery advertising that simultaneously maximizes impressions for new orders and balances supply distribution, and introduces a fast alternating local‑search algorithm (BOLS) that outperforms popular multi‑objective evolutionary algorithms and the commercial solver Gurobi in extensive experiments.
SPLAM (Sub‑Path Linear Approximation Model) accelerates diffusion‑model image synthesis by linearly approximating short sub‑paths of the probability‑flow ODE, allowing high‑quality generation in as few as four steps, outperforming prior fast‑sampling methods on COCO benchmarks and being deployed in Alibaba Mama’s recommendation system.
Alibaba’s advertising team introduces semantic‑aware contrastive learning to pre‑train multimodal image‑text embeddings, integrates them via SimTier and MAKE into ID‑based CTR models, achieving up to 6.9% lift in Taobao display ad click‑through rates and improving long‑tail item performance.
The paper presents AIGB, a generative auto‑bidding framework that replaces reinforcement‑learning with a conditional diffusion model to generate optimal bidding trajectories, and demonstrates through offline benchmarks and Alibaba’s online A/B tests that it consistently outperforms RL baselines, boosting buy count, GMV, and ROI while maintaining low latency.
