Lifu Huang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Lifu Huang has a total of 143 co-authors. Here are the top co-authors by number of shared publications:
Qifan Wang
7 shared publications- Mohammad Beigi
University of California Davis
5 shared publications Barry Menglong Yao
Virginia Tech ♣ UCLA ♢ Meta AI
3 shared publications- Jin-Hee Cho
Virginia Tech, Arlington, VA, USA
3 shared publications Ming Jin
3 shared publications- Alireza Shojaei
Virginia Tech, USA
2 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 15676-15685
Open-Ended Video Game Glitch Detection with Agentic Reasoning and Temporal Grounding
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 15676-15685
Open AccessOpen-Ended Video Game Glitch Detection with Agentic Reasoning and Temporal Grounding
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 6069-6078
GLANCE: A Global–Local Coordination Multi-Agent Framework for Music-Grounded Non-Linear Video Editing
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 6069-6078
Open AccessGLANCE: A Global–Local Coordination Multi-Agent Framework for Music-Grounded Non-Linear Video Editing
- Open Access
Foundation Models for Autonomous Robots in Unstructured Environments: Current State-of-the-Art, Challenges, and Future Pathways
Automating activities in unstructured environments, such as construction sites, has been challenging due to unpredictable events, limiting robot adoption compared to structured settings. Recently, pre-trained foundation models, particularly Large Language Models (LLMs), have shown promise in addressing this challenge through superior generalization capabilities. This study employed a multi-dimensional method that systematically reviews the field from different perspectives of foundation models in robotics and unstructured environments, and synthesizes them with deliberative acting theory. The findings revealed that LLMs’ linguistic capabilities are primarily used to improve perception and human–robot interactions in robotic tasks, while applications in project management, safety, and natural hazard detection are the most utilized applications of foundation models in unstructured environments. Our synthesis shows an empirical gap in the field where fewer identified studies within unstructured environments validated their foundation model applications using physically deployed robots. We positioned the current state-of-the-art on a five-level automation scale of conditional automation. These findings inform future scenarios, challenges, and solutions toward autonomous safe unstructured environments. Our study serves as a benchmark to track our progress toward that future.
Open AccessFoundation Models for Autonomous Robots in Unstructured Environments: Current State-of-the-Art, Challenges, and Future Pathways
Automating activities in unstructured environments, such as construction sites, has been challenging due to unpredictable events, limiting robot adoption compared to structured settings. Recently, pre-trained foundation models, particularly Large Language Models (LLMs), have shown promise in addressing this challenge through superior generalization capabilities. This study employed a multi-dimensional method that systematically reviews the field from different perspectives of foundation models in robotics and unstructured environments, and synthesizes them with deliberative acting theory. The findings revealed that LLMs’ linguistic capabilities are primarily used to improve perception and human–robot interactions in robotic tasks, while applications in project management, safety, and natural hazard detection are the most utilized applications of foundation models in unstructured environments. Our synthesis shows an empirical gap in the field where fewer identified studies within unstructured environments validated their foundation model applications using physically deployed robots. We positioned the current state-of-the-art on a five-level automation scale of conditional automation. These findings inform future scenarios, challenges, and solutions toward autonomous safe unstructured environments. Our study serves as a benchmark to track our progress toward that future.
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-25
From Vulnerable to Resilient: Examining Parent and Teen Perceptions on How to Respond to Unwanted Cybergrooming Advances
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-25
Open AccessFrom Vulnerable to Resilient: Examining Parent and Teen Perceptions on How to Respond to Unwanted Cybergrooming Advances
- Open Access
IR$^3$: Contrastive Inverse Reinforcement Learning for Interpretable Detection and Mitigation of Reward Hacking
Reinforcement Learning from Human Feedback (RLHF) enables powerful LLM alignment but can introduce reward hacking - models exploit spurious correlations in proxy rewards without genuine alignment. Compounding this, the objectives internalized during RLHF remain opaque, making hacking behaviors difficult to detect or correct. We introduce IR3 (Interpretable Reward Reconstruction and Rectification), a framework that reverse-engineers, interprets, and surgically repairs the implicit objectives driving RLHF-tuned models. We propose Contrastive Inverse Reinforcement Learning (C-IRL), which reconstructs the implicit reward function by contrasting paired responses from post-alignment and baseline policies to explain behavioral shifts during RLHF. We then decompose the reconstructed reward via sparse autoencoders into interpretable features, enabling identification of hacking signatures through contribution analysis. Finally, we propose mitigation strategies - clean reward optimization, adversarial shaping, constrained optimization, and feature-guided distillation - that target problematic features while preserving beneficial alignment. Experiments across multiple reward model configurations show that IR3 achieves 0.89 correlation with ground-truth rewards, identifies hacking features with over 90% precision, and significantly reduces hacking behaviors while maintaining capabilities within 3% of the original model.
Open AccessIR$^3$: Contrastive Inverse Reinforcement Learning for Interpretable Detection and Mitigation of Reward Hacking
Reinforcement Learning from Human Feedback (RLHF) enables powerful LLM alignment but can introduce reward hacking - models exploit spurious correlations in proxy rewards without genuine alignment. Compounding this, the objectives internalized during RLHF remain opaque, making hacking behaviors difficult to detect or correct. We introduce IR3 (Interpretable Reward Reconstruction and Rectification), a framework that reverse-engineers, interprets, and surgically repairs the implicit objectives driving RLHF-tuned models. We propose Contrastive Inverse Reinforcement Learning (C-IRL), which reconstructs the implicit reward function by contrasting paired responses from post-alignment and baseline policies to explain behavioral shifts during RLHF. We then decompose the reconstructed reward via sparse autoencoders into interpretable features, enabling identification of hacking signatures through contribution analysis. Finally, we propose mitigation strategies - clean reward optimization, adversarial shaping, constrained optimization, and feature-guided distillation - that target problematic features while preserving beneficial alignment. Experiments across multiple reward model configurations show that IR3 achieves 0.89 correlation with ground-truth rewards, identifies hacking features with over 90% precision, and significantly reduces hacking behaviors while maintaining capabilities within 3% of the original model.
- Open Access
Adversarial Reward Auditing for Active Detection and Mitigation of Reward Hacking
Reinforcement Learning from Human Feedback (RLHF) remains vulnerable to reward hacking, where models exploit spurious correlations in learned reward models to achieve high scores while violating human intent. Existing mitigations rely on static defenses that cannot adapt to novel exploitation strategies. We propose Adversarial Reward Auditing (ARA), a framework that reconceptualizes reward hacking as a dynamic, competitive game. ARA operates in two stages: first, a Hacker policy discovers reward model vulnerabilities while an Auditor learns to detect exploitation from latent representations; second, Auditor-Guided RLHF (AG-RLHF) gates reward signals to penalize detected hacking, transforming reward hacking from an unobservable failure into a measurable, controllable signal. Experiments across three hacking scenarios demonstrate that ARA achieves the best alignment-utility tradeoff among all baselines: reducing sycophancy to near-SFT levels while improving helpfulness, decreasing verbosity while achieving the highest ROUGE-L, and suppressing code gaming while improving Pass@1. Beyond single-domain evaluation, we show that reward hacking, detection, and mitigation all generalize across domains -- a Hacker trained on code gaming exhibits increased sycophancy despite no reward for this behavior, and an Auditor trained on one domain effectively suppresses exploitation in others, enabling efficient multi-domain defense with a single model.
Open AccessAdversarial Reward Auditing for Active Detection and Mitigation of Reward Hacking
Reinforcement Learning from Human Feedback (RLHF) remains vulnerable to reward hacking, where models exploit spurious correlations in learned reward models to achieve high scores while violating human intent. Existing mitigations rely on static defenses that cannot adapt to novel exploitation strategies. We propose Adversarial Reward Auditing (ARA), a framework that reconceptualizes reward hacking as a dynamic, competitive game. ARA operates in two stages: first, a Hacker policy discovers reward model vulnerabilities while an Auditor learns to detect exploitation from latent representations; second, Auditor-Guided RLHF (AG-RLHF) gates reward signals to penalize detected hacking, transforming reward hacking from an unobservable failure into a measurable, controllable signal. Experiments across three hacking scenarios demonstrate that ARA achieves the best alignment-utility tradeoff among all baselines: reducing sycophancy to near-SFT levels while improving helpfulness, decreasing verbosity while achieving the highest ROUGE-L, and suppressing code gaming while improving Pass@1. Beyond single-domain evaluation, we show that reward hacking, detection, and mitigation all generalize across domains -- a Hacker trained on code gaming exhibits increased sycophancy despite no reward for this behavior, and an Auditor trained on one domain effectively suppresses exploitation in others, enabling efficient multi-domain defense with a single model.
- Open Access
Zero-shot adaptable task planning for autonomous construction robots: a comparative study of lightweight single and multi-AI agent systems
Robots are expected to play a major role in the future construction industry but face challenges due to high costs and difficulty adapting to dynamic tasks. This study explores the potential of foundation models to enhance the adaptability and generalizability of task planning in construction robots. Four models are proposed and implemented using lightweight, open-source large language models (LLMs) and vision language models (VLMs). These models include one single agent and three multi-agent teams that collaborate to create robot action plans. The models are evaluated across three construction roles: Painter, Safety Inspector, and Floor Tiling. Results show that the four-agent team outperforms the state-of-the-art GPT-4o in most metrics while being ten times more cost-effective. Additionally, teams with three and four agents demonstrate the improved generalizability. By discussing how agent behaviors influence outputs, this study enhances the understanding of AI teams and supports future research in diverse unstructured environments beyond construction.
Open AccessZero-shot adaptable task planning for autonomous construction robots: a comparative study of lightweight single and multi-AI agent systems
Robots are expected to play a major role in the future construction industry but face challenges due to high costs and difficulty adapting to dynamic tasks. This study explores the potential of foundation models to enhance the adaptability and generalizability of task planning in construction robots. Four models are proposed and implemented using lightweight, open-source large language models (LLMs) and vision language models (VLMs). These models include one single agent and three multi-agent teams that collaborate to create robot action plans. The models are evaluated across three construction roles: Painter, Safety Inspector, and Floor Tiling. Results show that the four-agent team outperforms the state-of-the-art GPT-4o in most metrics while being ten times more cost-effective. Additionally, teams with three and four agents demonstrate the improved generalizability. By discussing how agent behaviors influence outputs, this study enhances the understanding of AI teams and supports future research in diverse unstructured environments beyond construction.
- Open Access
Navigating Ideation Space: Decomposed Conceptual Representations for Positioning Scientific Ideas
Scientific discovery is a cumulative process and requires new ideas to be situated within an ever-expanding landscape of existing knowledge. An emerging and critical challenge is how to identify conceptually relevant prior work from rapidly growing literature, and assess how a new idea differentiates from existing research. Current embedding approaches typically conflate distinct conceptual aspects into single representations and cannot support fine-grained literature retrieval; meanwhile, LLM-based evaluators are subject to sycophancy biases, failing to provide discriminative novelty assessment. To tackle these challenges, we introduce the Ideation Space, a structured representation that decomposes scientific knowledge into three distinct dimensions, i.e., research problem, methodology, and core findings, each learned through contrastive training. This framework enables principled measurement of conceptual distance between ideas, and modeling of ideation transitions that capture the logical connections within a proposed idea. Building upon this representation, we propose a Hierarchical Sub-Space Retrieval framework for efficient, targeted literature retrieval, and a Decomposed Novelty Assessment algorithm that identifies which aspects of an idea are novel. Extensive experiments demonstrate substantial improvements, where our approach achieves Recall@30 of 0.329 (16.7% over baselines), our ideation transition retrieval reaches Hit Rate@30 of 0.643, and novelty assessment attains 0.37 correlation with expert judgments. In summary, our work provides a promising paradigm for future research on accelerating and evaluating scientific discovery.
Open AccessNavigating Ideation Space: Decomposed Conceptual Representations for Positioning Scientific Ideas
Scientific discovery is a cumulative process and requires new ideas to be situated within an ever-expanding landscape of existing knowledge. An emerging and critical challenge is how to identify conceptually relevant prior work from rapidly growing literature, and assess how a new idea differentiates from existing research. Current embedding approaches typically conflate distinct conceptual aspects into single representations and cannot support fine-grained literature retrieval; meanwhile, LLM-based evaluators are subject to sycophancy biases, failing to provide discriminative novelty assessment. To tackle these challenges, we introduce the Ideation Space, a structured representation that decomposes scientific knowledge into three distinct dimensions, i.e., research problem, methodology, and core findings, each learned through contrastive training. This framework enables principled measurement of conceptual distance between ideas, and modeling of ideation transitions that capture the logical connections within a proposed idea. Building upon this representation, we propose a Hierarchical Sub-Space Retrieval framework for efficient, targeted literature retrieval, and a Decomposed Novelty Assessment algorithm that identifies which aspects of an idea are novel. Extensive experiments demonstrate substantial improvements, where our approach achieves Recall@30 of 0.329 (16.7% over baselines), our ideation transition retrieval reaches Hit Rate@30 of 0.643, and novelty assessment attains 0.37 correlation with expert judgments. In summary, our work provides a promising paradigm for future research on accelerating and evaluating scientific discovery.
- Open Access
How Do Large Language Models Learn Concepts During Continual Pre-Training?
Human beings primarily understand the world through concepts (e.g., dog), abstract mental representations that structure perception, reasoning, and learning. However, how large language models (LLMs) acquire, retain, and forget such concepts during continual pretraining remains poorly understood. In this work, we study how individual concepts are acquired and forgotten, as well as how multiple concepts interact through interference and synergy. We link these behavioral dynamics to LLMs' internal Concept Circuits, computational subgraphs associated with specific concepts, and incorporate Graph Metrics to characterize circuit structure. Our analysis reveals: (1) LLMs concept circuits provide a non-trivial, statistically significant signal of concept learning and forgetting; (2) Concept circuits exhibit a stage-wise temporal pattern during continual pretraining, with an early increase followed by gradual decrease and stabilization; (3) concepts with larger learning gains tend to exhibit greater forgetting under subsequent training; (4) semantically similar concepts induce stronger interference than weakly related ones; (5) conceptual knowledge differs in their transferability, with some significantly facilitating the learning of others. Together, our findings offer a circuit-level view of concept learning dynamics and inform the design of more interpretable and robust concept-aware training strategies for LLMs.
Open AccessHow Do Large Language Models Learn Concepts During Continual Pre-Training?
Human beings primarily understand the world through concepts (e.g., dog), abstract mental representations that structure perception, reasoning, and learning. However, how large language models (LLMs) acquire, retain, and forget such concepts during continual pretraining remains poorly understood. In this work, we study how individual concepts are acquired and forgotten, as well as how multiple concepts interact through interference and synergy. We link these behavioral dynamics to LLMs' internal Concept Circuits, computational subgraphs associated with specific concepts, and incorporate Graph Metrics to characterize circuit structure. Our analysis reveals: (1) LLMs concept circuits provide a non-trivial, statistically significant signal of concept learning and forgetting; (2) Concept circuits exhibit a stage-wise temporal pattern during continual pretraining, with an early increase followed by gradual decrease and stabilization; (3) concepts with larger learning gains tend to exhibit greater forgetting under subsequent training; (4) semantically similar concepts induce stronger interference than weakly related ones; (5) conceptual knowledge differs in their transferability, with some significantly facilitating the learning of others. Together, our findings offer a circuit-level view of concept learning dynamics and inform the design of more interpretable and robust concept-aware training strategies for LLMs.
- 19 December 2025
- Published by Elsevier in Information Fusion
- Vol. 130
Survey of uncertainty estimation in LLMs - Sources, methods, applications, and challenges
- 19 December 2025
- Published by Elsevier in Information Fusion
- Vol. 130
Survey of uncertainty estimation in LLMs - Sources, methods, applications, and challenges