Enhancing Recommendation Models with Scaling Law via HCNet and MemoNet: A Memory‑Based Feature‑Combination Approach

This article presents a memory‑driven architecture (HCNet and MemoNet) that equips recommendation models with scaling‑law characteristics by storing and retrieving arbitrary feature‑combination embeddings, evaluates multi‑hash codebooks, memory‑restoring strategies, key‑feature selection, and demonstrates significant offline and online performance gains.

feature interactionlarge language modelsmemory networks

0 likes · 15 min read

Enhancing Recommendation Models with Scaling Law via HCNet and MemoNet: A Memory‑Based Feature‑Combination Approach

DataFunTalk

Dec 1, 2022 · Artificial Intelligence

Advances and Challenges in Controllable Text Generation with Pretrained Language Models

This report reviews the background, recent research progress, practical applications, and future directions of controllable text generation using transformer‑based pretrained language models, highlighting methods such as decoding strategies, prompt learning, memory networks, continual learning, contrastive training, and knowledge integration.

continual learningcontrastive trainingcontrollable text generation

0 likes · 13 min read

Advances and Challenges in Controllable Text Generation with Pretrained Language Models

Alibaba Cloud Developer

Nov 7, 2019 · Artificial Intelligence

Boosting Task-Oriented Dialogue with Heterogeneous Memory Networks

This paper introduces Heterogeneous Memory Networks (HMNs), combining context‑free and context‑aware memory modules to jointly process user queries, dialogue history, and knowledge bases, achieving state‑of‑the‑art performance on three task‑oriented dialogue datasets in both BLEU and F1 metrics.

Dialogue Systemsknowledge integrationmemory networks

0 likes · 17 min read

Boosting Task-Oriented Dialogue with Heterogeneous Memory Networks

Alibaba Cloud Developer

Aug 14, 2019 · Artificial Intelligence

How MIMN+UIC Breaks the Long-Sequence Barrier in Real-Time CTR Prediction

This article presents a co-designed algorithm‑system solution—MIMN and an independent UIC module—that enables ultra‑long user behavior modeling for click‑through rate prediction, delivering significant offline AUC gains and online CTR/RPM improvements in Alibaba's display advertising platform.

CTR predictionDeep LearningRecommendation Systems

0 likes · 12 min read

How MIMN+UIC Breaks the Long-Sequence Barrier in Real-Time CTR Prediction

Tencent Cloud Developer

Dec 21, 2018 · Artificial Intelligence

Tencent Xiaowei Conversational AI Platform: Architecture, Models, and Applications

Tencent Xiaowei is an open, easy‑to‑integrate conversational AI platform that combines NLU, dialogue management and generation, supports multi‑turn context via Memory Networks, uses bidirectional RNN and CNN‑based intent classifiers, and powers smart speakers, TVs and customer‑service bots by leveraging Tencent’s rich content ecosystem.

Conversational AIDialogue SystemsNLP

0 likes · 11 min read

Tencent Xiaowei Conversational AI Platform: Architecture, Models, and Applications