# Reward Finetuning for Faster and More Accurate Unsupervised Object Discovery

arXiv: https://arxiv.org/abs/2310.19080 · HTML: https://mlsanity.com/paper/2310.19080/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 35 of its 53 references that are in the map.
- First posted on arXiv 29 Oct 2023.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Reward modeling and RLHF](https://mlsanity.com/alignment/reward-modeling-and-rlhf/)

## Builds on (most cited first, 30 of 35)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [PaLM: Scaling Language Modeling with Pathways](https://mlsanity.com/paper/2204.02311/) (2022)
- [nuScenes: A multimodal dataset for autonomous driving](https://mlsanity.com/paper/1903.11027/) (2019)
- [Zero-Shot Text-to-Image Generation](https://mlsanity.com/paper/2102.12092/) (2021)
- [On the Opportunities and Risks of Foundation Models](https://mlsanity.com/paper/2108.07258/) (2021)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [Emergent Abilities of Large Language Models](https://mlsanity.com/paper/2206.07682/) (2022)
- [PointPillars: Fast Encoders for Object Detection from Point Clouds](https://mlsanity.com/paper/1812.05784/) (2018)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [PointRCNN: 3D Object Proposal Generation and Detection from Point Cloud](https://mlsanity.com/paper/1812.04244/) (2018)
- [PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection](https://mlsanity.com/paper/1912.13192/) (2019)
- [Multi-View 3D Object Detection Network for Autonomous Driving](https://mlsanity.com/paper/1611.07759/) (2016)
- [RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment](https://mlsanity.com/paper/2304.06767/) (2023)
- [Deep Hough Voting for 3D Object Detection in Point Clouds](https://mlsanity.com/paper/1904.09664/) (2019)
- [3DSSD: Point-based 3D Single Stage Object Detector](https://mlsanity.com/paper/2002.10187/) (2020)
- [Curriculum Learning for Reinforcement Learning Domains: A Framework and Survey](https://mlsanity.com/paper/2003.04960/) (2020)
- [STD: Sparse-to-Dense 3D Object Detector for Point Cloud](https://mlsanity.com/paper/1907.10471/) (2019)
- [Quark: Controllable Text Generation with Reinforced Unlearning](https://mlsanity.com/paper/2205.13636/) (2022)
- [Self-Supervised Transformers for Unsupervised Object Discovery using Normalized Cut](https://mlsanity.com/paper/2202.11539/) (2022)
- [Train in Germany, Test in The USA: Making 3D Object Detectors Generalize](https://mlsanity.com/paper/2005.08139/) (2020)
- [Learning Features by Watching Objects Move](https://mlsanity.com/paper/1612.06370/) (2016)
- [Imitation Learning from Imperfect Demonstration](https://mlsanity.com/paper/1901.09387/) (2019)
- [Ithaca365: Dataset and Driving Perception under Repeated and Challenging Weather Conditions](https://mlsanity.com/paper/2208.01166/) (2022)
- [Learning to Detect Mobile Objects from LiDAR Scans Without Labels](https://mlsanity.com/paper/2203.15882/) (2022)
- [Tuning computer vision models with task rewards](https://mlsanity.com/paper/2302.08242/) (2023)
- [Driven to Distraction: Self-Supervised Distractor Learning for Robust Monocular Visual Odometry in Urban Environments](https://mlsanity.com/paper/1711.06623/) (2017)

## Built on by (most cited first, 14 of 14)

- [DriveDPO: Policy Learning via Safety DPO For End-to-End Autonomous Driving](https://mlsanity.com/paper/2509.17940/) (2025)
- [UNION: Unsupervised 3D Object Detection using Object Appearance-based Pseudo-Classes](https://mlsanity.com/paper/2405.15688/) (2024)
- [Harnessing Uncertainty-aware Bounding Boxes for Unsupervised 3D Object Detection](https://mlsanity.com/paper/2408.00619/) (2024)
- [DiffuBox: Refining 3D Object Detection with Point Diffusion](https://mlsanity.com/paper/2405.16034/) (2024)
- [Learning to Detect Objects from Multi-Agent LiDAR Scans without Manual Labels](https://mlsanity.com/paper/2503.08421/) (2025)
- [SP3D: Boosting Sparsely-Supervised 3D Object Detection via Accurate Cross-Modal Semantic Prompts](https://mlsanity.com/paper/2503.06467/) (2025)
- [Learning 3D Perception from Others' Predictions](https://mlsanity.com/paper/2410.02646/) (2024)
- [Label-Efficient 3D Object Detection For Road-Side Units](https://mlsanity.com/paper/2404.06256/) (2024)
- [Vision-Language Guidance for LiDAR-based Unsupervised 3D Object Detection](https://mlsanity.com/paper/2408.03790/) (2024)
- [MonoSOWA: Scalable monocular 3D Object detector Without human Annotations](https://mlsanity.com/paper/2501.09481/) (2025)
- [Seg2Box: 3D Object Detection by Point-Wise Semantics Supervision](https://mlsanity.com/paper/2503.16811/) (2025)
- [VESPA: Towards un(Human)supervised Open-World Pointcloud Labeling for Autonomous Driving](https://mlsanity.com/paper/2507.20397/) (2025)
- [xMOD: Cross-Modal Distillation for 2D/3D Multi-Object Discovery from 2D motion](https://mlsanity.com/paper/2503.15022/) (2025)
- [When the City Teaches the Car: Label-Free 3D Perception from Infrastructure](https://mlsanity.com/paper/2603.16742/) (2026)

## Abstract (arXiv)

Recent advances in machine learning have shown that Reinforcement Learning from Human Feedback (RLHF) can improve machine learning models and align them with human preferences. Although very successful for Large Language Models (LLMs), these advancements have not had a comparable impact in research for autonomous vehicles -- where alignment with human expectations can be imperative. In this paper, we propose to adapt similar RL-based methods to unsupervised object discovery, i.e. learning to detect objects from LiDAR points without any training labels. Instead of labels, we use simple heuristics to mimic human feedback. More explicitly, we combine multiple heuristics into a simple reward function that positively correlates its score with bounding box accuracy, i.e., boxes containing objects are scored higher than those without. We start from the detector's own predictions to explore the space and reinforce boxes with high rewards through gradient updates. Empirically, we demonstrate that our approach is not only more accurate, but also orders of magnitudes faster to train compared to prior works on object discovery.