# Latent-Aligned Reasoning for Multimodal Recommendation

arXiv: https://arxiv.org/abs/2609.04645 · HTML: https://mlsanity.com/paper/2609.04645/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 24 of its 28 references that are in the map.
- First posted on arXiv 4 Sep 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 24 of 24)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [Self-Attentive Sequential Recommendation](https://mlsanity.com/paper/1808.09781/) (2018)
- [LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation](https://mlsanity.com/paper/2002.02126/) (2020)
- [DeepFM: A Factorization-Machine based Neural Network for CTR Prediction](https://mlsanity.com/paper/1703.04247/) (2017)
- [Ups and Downs: Modeling the Visual Evolution of Fashion Trends with One-Class Collaborative Filtering](https://mlsanity.com/paper/1602.01585/) (2016)
- [Imagine while Reasoning in Space: Multimodal Visualization-of-Thought](https://mlsanity.com/paper/2501.07542/) (2025)
- [LLMRec: Large Language Models with Graph Augmentation for Recommendation](https://mlsanity.com/paper/2311.00423/) (2023)
- [Representation Learning with Large Language Models for Recommendation](https://mlsanity.com/paper/2310.15950/) (2023)
- [VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback](https://mlsanity.com/paper/1510.01784/) (2015)
- [Multimodal Intelligence: Representation Learning, Information Fusion, and Applications](https://mlsanity.com/paper/1911.03977/) (2019)
- [Mining Latent Structures for Multimedia Recommendation](https://mlsanity.com/paper/2104.09036/) (2021)
- [Bootstrap Latent Representations for Multi-modal Recommendation](https://mlsanity.com/paper/2207.05969/) (2022)
- [A Tale of Two Graphs: Freezing and Denoising Graph Structures for Multimodal Recommendation](https://mlsanity.com/paper/2211.06924/) (2022)
- [CapArena: Benchmarking and Analyzing Detailed Image Captioning in the LLM Era](https://mlsanity.com/paper/2503.12329/) (2025)
- [Large Scale Product Graph Construction for Recommendation in E-commerce](https://mlsanity.com/paper/2010.05525/) (2020)
- [Visual question answering: from early developments to recent advances -- a survey](https://mlsanity.com/paper/2501.03939/) (2025)
- [Why Do We Click: Visual Impression-aware News Recommendation](https://mlsanity.com/paper/2109.12651/) (2021)
- [Why Not Together? A Multiple-Round Recommender System for Queries and Items](https://mlsanity.com/paper/2412.10787/) (2024)

## Abstract (arXiv)

Multimodal Vision-Language Models (VLMs) have demonstrated remarkable capabilities in cross-modal understanding, yet a fundamental challenge persists when applying them to recommendation: as representations propagate through multi-step reasoning, both visual and textual signals progressively attenuate - a phenomenon we term cross-modal dilution. To address this, we propose LARK (Latent-Aligned Reasoning frameworK), a two-stage latent reasoning framework with complementary alignment mechanisms within a single VLM. In the first stage, learnable latent tokens are interleaved with multi-step chain-of-thought (CoT) reasoning and explicitly aligned with a frozen vision encoder, serving as visual checkpoints that preserve perceptual details throughout the reasoning chain. In the second stage, the latent representations are projected via a bridge MLP and trained with item-to-item contrastive learning; to prevent the reasoning semantics from fading, intermediate features are aligned with the CoT hidden states from the first stage, anchoring the final embeddings to the model's own reasoning output. Experiments on three public benchmarks and one industrial dataset show that LARK achieves state-of-the-art performance across multiple recommendation architectures, with controlled ablations confirming the distinct contribution of each component.