OSCAR: Optimization-Steered Agentic Planning for Composed Image Retrieval

• URL: http://arxiv.org/abs/2602.08603v1
• Date: Mon, 09 Feb 2026 12:44:56 GMT
• Title: OSCAR: Optimization-Steered Agentic Planning for Composed Image Retrieval
• Authors: Teng Wang, Rong Shan, Jianghao Lin, Junjie Wu, Tianyi Xu, Jianping Zhang, Wenteng Chen, Changwang Zhang, Zhaoxiang Wang, Weinan Zhang, Jun Wang,
• Abstract summary: We propose OSCAR, an optimization-steered agentic planning framework for composed image retrieval.<n>We are the first to reformulate agentic CIR from a search process into a principled trajectory optimization problem.<n>In the offline phase, we model CIR via atomic retrieval selection and composition as a two-stage mixed-integer programming problem.<n>These trajectories are then stored in a golden library to serve as in-context demonstrations for online steering of VLM planner.
• Score: 33.823055061609125
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Composed image retrieval (CIR) requires complex reasoning over heterogeneous visual and textual constraints. Existing approaches largely fall into two paradigms: unified embedding retrieval, which suffers from single-model myopia, and heuristic agentic retrieval, which is limited by suboptimal, trial-and-error orchestration. To this end, we propose OSCAR, an optimization-steered agentic planning framework for composed image retrieval. We are the first to reformulate agentic CIR from a heuristic search process into a principled trajectory optimization problem. Instead of relying on heuristic trial-and-error exploration, OSCAR employs a novel offline-online paradigm. In the offline phase, we model CIR via atomic retrieval selection and composition as a two-stage mixed-integer programming problem, mathematically deriving optimal trajectories that maximize ground-truth coverage for training samples via rigorous boolean set operations. These trajectories are then stored in a golden library to serve as in-context demonstrations for online steering of VLM planner at online inference time. Extensive experiments on three public benchmarks and a private industrial benchmark show that OSCAR consistently outperforms SOTA baselines. Notably, it achieves superior performance using only 10% of training data, demonstrating strong generalization of planning logic rather than dataset-specific memorization.

Related papers

• Rethinking Autoregressive Models for Lossless Image Compression via Hierarchical Parallelism and Progressive Adaptation [75.58269386927076]
  Autoregressive (AR) models are often dismissed as impractical due to prohibitive computational cost.<n>This work re-thinks this paradigm, introducing a framework built on hierarchical parallelism and progressive adaptation.<n> Experiments on diverse datasets (natural, satellite, medical) validate that our method achieves new state-of-the-art compression.
  arXiv  Detail & Related papers  (2025-11-14T06:27:58Z)
• ATLAS: Actor-Critic Task-Completion with Look-ahead Action Simulation [28.54052846801967]
  ATLAS is a memory-augmented agent that makes plans grounded in a model of the environment by simulating the consequences of those actions in cognitive space.<n>On the WebArena-Lite Benchmark, we achieve a 63% success rate compared to 53.9% success rate for the previously published state-of-the-art.
  arXiv  Detail & Related papers  (2025-10-26T16:03:39Z)
• VAR: Visual Attention Reasoning via Structured Search and Backtracking [49.427842994857635]
  We introduce Visual Attention Reasoning, a framework that recasts grounded reasoning as a structured search.<n> VAR decomposes the reasoning process into two key stages: traceable evidence grounding and search-based chain-of-thought.<n>We show that our 7B model, VAR-7B, sets a new state-of-the-art on a comprehensive suite of hallucination and safety benchmarks.
  arXiv  Detail & Related papers  (2025-10-21T13:18:44Z)
• TaoSR1: The Thinking Model for E-commerce Relevance Search [15.137901457184839]
  BERT-based models excel at semantic matching but lack complex reasoning capabilities.<n>We propose a framework to directly deploy Large Language Models for this task, addressing key challenges: Chain-of-Thought (CoT) error accumulation, discriminative hallucination, and deployment feasibility.<n>Our framework, TaoSR1, involves three stages: (1) Supervised Fine-Tuning (SFT) with CoT to instill reasoning; (2) Offline sampling with a pass@N strategy and Direct Preference Optimization (DPO) to improve generation quality; and (3) Difficulty-based dynamic sampling with Group Relative Policy Optimization (GRPO)
  arXiv  Detail & Related papers  (2025-08-17T13:48:48Z)
• You Only Train Once [11.97836331714694]
  You Only Train Once (YOTO) contributes to limiting training to one shot for the latter aspect of losses selection and weighting.<n>We leverage the differentiability of the composite loss formulation which is widely used for optimizing multiple empirical losses simultaneously.<n>We show that YOTO consistently outperforms the best grid-search model on unseen test data.
  arXiv  Detail & Related papers  (2025-06-04T18:04:58Z)
• SimpleDeepSearcher: Deep Information Seeking via Web-Powered Reasoning Trajectory Synthesis [94.33978856270268]
  Retrieval-augmented generation (RAG) systems have advanced large language models (LLMs) in complex deep search scenarios.<n>Existing approaches face critical limitations that lack high-quality training trajectories and suffer from distributional mismatches.<n>This paper introduces SimpleDeepSearcher, a framework that bridges the gap through strategic data engineering rather than complex training paradigms.
  arXiv  Detail & Related papers  (2025-05-22T16:05:02Z)
• Pseudo-triplet Guided Few-shot Composed Image Retrieval [20.040511832864503]
  Composed Image Retrieval (CIR) is a challenging task that aims to retrieve the target image with a multimodal query. We propose a novel two-stage pseudo triplet guided few-shot CIR scheme, dubbed PTG-FSCIR. In the first stage, we propose an attentive masking and captioning-based pseudo triplet generation method, to construct pseudo triplets from pure image data. In the second stage, we propose a challenging triplet-based CIR fine-tuning method, where we design a pseudo modification text-based sample challenging score estimation strategy.
  arXiv  Detail & Related papers  (2024-07-08T14:53:07Z)
• Maximize to Explore: One Objective Function Fusing Estimation, Planning, and Exploration [87.53543137162488]
  We propose an easy-to-implement online reinforcement learning (online RL) framework called textttMEX. textttMEX integrates estimation and planning components while balancing exploration exploitation automatically. It can outperform baselines by a stable margin in various MuJoCo environments with sparse rewards.
  arXiv  Detail & Related papers  (2023-05-29T17:25:26Z)
• End-to-End Meta-Bayesian Optimisation with Transformer Neural Processes [52.818579746354665]
  This paper proposes the first end-to-end differentiable meta-BO framework that generalises neural processes to learn acquisition functions via transformer architectures. We enable this end-to-end framework with reinforcement learning (RL) to tackle the lack of labelled acquisition data.
  arXiv  Detail & Related papers  (2023-05-25T10:58:46Z)
• Don't Be So Sure! Boosting ASR Decoding via Confidence Relaxation [7.056222499095849]
  beam search seeks the transcript with the greatest likelihood computed using the predicted distribution. We show that recently proposed Self-Supervised Learning (SSL)-based ASR models tend to yield exceptionally confident predictions. We propose a decoding procedure that improves the performance of fine-tuned ASR models.
  arXiv  Detail & Related papers  (2022-12-27T06:42:26Z)
• TransPath: Learning Heuristics For Grid-Based Pathfinding via Transformers [64.88759709443819]
  We suggest learning the instance-dependent proxies that are supposed to notably increase the efficiency of the search. The first proxy we suggest to learn is the correction factor, i.e. the ratio between the instance independent cost-to-go estimate and the perfect one. The second proxy is the path probability, which indicates how likely the grid cell is lying on the shortest path.
  arXiv  Detail & Related papers  (2022-12-22T14:26:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.