New arXiv work shows they can be poisoned with under 1% synthetic interactions, significantly boosting malicious targets via decoding-aligned prefixes. Worth considering if you build or evaluate…

#GenerativeAI #PromptInjection #RecommenderSystems #AIResearch
https://arxiv.org/abs/2610.14487

New research introduces STATIC, a method that flattens prefix trees into a compressed sparse row matrix to enable efficient constrained decoding for LLM-based generative retrieval on TPUs and GPUs. The approach…

#AIInfrastructure #GPUs #LLMs #RecommenderSystems
https://arxiv.org/abs/2602.22647

CutBCE offers a fused, hardware-accelerated BCE loss and gradient operator in JAX and Pallas, tackling OOM issues in large-vocabulary multi-label recommendation without changing the math. Useful for scaling sequential…

#AI #MachineLearning #RecommenderSystems #JAX
https://arxiv.org/abs/2610.05559

Yandex’s new Sona recommender blends dual attention to weigh both historic and fresh user signals—boosting relevance like never before. Curious how it balances past & recent events? Dive into the research! #YandexSona #DualAttention #RecommenderSystems

🔗 aidailypost.com/news/yandex-...

New paper "ImbalancE" proposes an inference-time latent search method to fix degree imbalance in knowledge graph link prediction, targeting hard cases where the target entity has far fewer connections than…

#AI #KnowledgeGraphs #LinkPrediction #RecommenderSystems
https://arxiv.org/abs/2609.36996

T-RoPE introduces a time-aware variant of Rotary Position Embedding for sequential generative recommendation, using timestamps instead of interaction indices to better capture elapsed time…

#RecommenderSystems #TransformerAI #SequentialRecommendation #GenerativeAI
https://arxiv.org/abs/2609.30576

Spotify’s Taste Profile lets Premium users fine-tune the AI behind Discover Weekly, Wrapped & more. #Spotify #TasteProfile #AI #RecommenderSystems #Streaming #MusicTech thedailytechfeed.com/spotify-lets...