arXiv subjects

Bo Peng

Publications and source records attributed to Bo Peng.

At least 127 records · Page 7Linked to original sources

arXiv2025

Dark Miner: Defend against undesirable generation for text-to-image diffusion models

Text-to-image diffusion models have been demonstrated with undesired generation due to unfiltered large-scale training data, such as sexual images and copyrights, necessitating the erasure of undesired concepts. Most existing methods focus on modifying the generation probabilities conditioned on the texts containing target concepts. However, they fail to guarantee the desired generation of texts unseen in the training phase, especially for the adversarial texts from malicious attacks. In this paper, we analyze the erasure task and point out that existing methods cannot guarantee the minimization of the total probabilities of undesired generation. To tackle this problem, we propose Dark Miner. It entails a recurring three-stage process that comprises mining, verifying, and circumventing. This method greedily mines embeddings with maximum generation probabilities of target concepts and more effectively reduces their generation. In the experiments, we evaluate its performance on the inappropriateness, object, and style concepts. Compared with the previous methods, our method achieves better erasure and defense results, especially under multiple adversarial attacks, while preserving the native generation capability of the models. Our code will be available on GitHub.

cs.CV↗

arXiv2025

Planning with Diffusion Models for Target-Oriented Dialogue Systems

Target-Oriented Dialogue (TOD) remains a significant challenge in the LLM era, where strategic dialogue planning is crucial for directing conversations toward specific targets. However, existing dialogue planning methods generate dialogue plans in a step-by-step sequential manner, and may suffer from compounding errors and myopic actions. To address these limitations, we introduce a novel dialogue planning framework, DiffTOD, which leverages diffusion models to enable non-sequential dialogue planning. DiffTOD formulates dialogue planning as a trajectory generation problem with conditional guidance, and leverages a diffusion language model to estimate the likelihood of the dialogue trajectory. To optimize the dialogue action strategies, DiffTOD introduces three tailored guidance mechanisms for different target types, offering flexible guidance toward diverse TOD targets at test time. Extensive experiments across three diverse TOD settings show that DiffTOD can effectively perform non-myopic lookahead exploration and optimize action strategies over a long horizon through non-sequential dialogue planning, and demonstrates strong flexibility across complex and diverse dialogue scenarios. Our code and data are accessible through https://github.com/ninglab/DiffTOD.

cs.CL↗

arXiv2025

Illuminating Hidden Pulsars: Scintillation-Enhanced Discovery of Two Binary Millisecond Pulsars in M13 with FAST

We conducted a sensitive acceleration search using Fast Fourier Transform (FFT) techniques on full-length and segmented data from 84 observations of the globular cluster M13 with the Five-hundred-meter Aperture Spherical radio Telescope (FAST). Employing a low detection threshold (2 $σ$) to maximize sensitivity to faint pulsars, here we report the discovery of two binary millisecond pulsars: J1641+3627G (M13G) and J1641+3627H (M13H). Both pulsars were detected during scintillation-brightened states, revealing systems that would otherwise remain undetected. For M13G, we obtained a phase-connected timing solution spanning 6.4 years, identifying it as a black widow system with an orbital period of 0.12 days hosting an extremely low-mass companion ($\sim 9.9\times 10^{-3}~{ M}_\odot$), though no eclipses were observed. M13H, however, shows significant apparent acceleration but was detected in only 2 of 84 observations; its extremely low detection rate currently prevents constraints on orbital parameters or classification.

astro-ph.HE↗

arXiv2025

CLIP-AGIQA: Boosting the Performance of AI-Generated Image Quality Assessment with CLIP

With the rapid development of generative technologies, AI-Generated Images (AIGIs) have been widely applied in various aspects of daily life. However, due to the immaturity of the technology, the quality of the generated images varies, so it is important to develop quality assessment techniques for the generated images. Although some models have been proposed to assess the quality of generated images, they are inadequate when faced with the ever-increasing and diverse categories of generated images. Consequently, the development of more advanced and effective models for evaluating the quality of generated images is urgently needed. Recent research has explored the significant potential of the visual language model CLIP in image quality assessment, finding that it performs well in evaluating the quality of natural images. However, its application to generated images has not been thoroughly investigated. In this paper, we build on this idea and further explore the potential of CLIP in evaluating the quality of generated images. We design CLIP-AGIQA, a CLIP-based regression model for quality assessment of generated images, leveraging rich visual and textual knowledge encapsulated in CLIP. Particularly, we implement multi-category learnable prompts to fully utilize the textual knowledge in CLIP for quality assessment. Extensive experiments on several generated image quality assessment benchmarks, including AGIQA-3K and AIGCIQA2023, demonstrate that CLIP-AGIQA outperforms existing IQA models, achieving excellent results in evaluating the quality of generated images.

cs.CV↗

arXiv2025

Friction-Aware Safety Locomotion for Wheeled-legged Robots using Vision Language Models and Reinforcement Learning

Controlling Wheeled-legged robots is challenging especially on slippery surfaces due to their dependence on continuous ground contact. Unlike quadrupeds or bipeds, which can leverage multiple fixed contact points for recovery, wheeled-legged robots are highly susceptible to slip, where even momentary loss of traction can result in irrecoverable instability. Anticipating ground physical properties such as friction before contact would allow proactive control adjustments, reducing slip risk. In this paper, we propose a friction-aware safety locomotion framework that integrates Vision-Language Models (VLMs) with a Reinforcement Learning (RL) policy. Our method employs a Retrieval-Augmented Generation (RAG) approach to estimate the Coefficient of Friction (CoF), which is then explicitly incorporated into the RL policy. This enables the robot to adapt its speed based on predicted friction conditions before contact. The framework is validated through experiments in both simulation and on a physical customized Wheeled Inverted Pendulum (WIP). Experimental results show that our approach successfully completes trajectory tracking tasks on slippery surfaces, whereas baseline methods relying solely on proprioceptive feedback fail. These findings highlight the importance and effectiveness of explicitly predicting and utilizing ground friction information for safe locomotion. They also point to a promising research direction in exploring the use of VLMs for estimating ground conditions, which remains a significant challenge for purely vision-based methods.

cs.RO↗

arXiv2025

Fine-structure Line Atlas for Multi-wavelength Extragalactic Study (FLAMES) III: [C II] as Tracer, Crisis of SFR, [O III]/[C II] at High-z, New Answers and New Questions

In the final paper of this series, we discuss new perspectives and challenges in the study of interstellar medium (ISM), leveraging comprehensive catalogs and physical insights presented in our previous papers. We focus on key questions of far-infrared (FIR) fine-structure lines (FSLs): their origins, diagnostic value, and implications of correlations. Our analysis reveals a strong dependence on elemental abundance, so that FSL/H$α$ traces metallicity, [N II]/[C II] traces N/O, and $\sim$80% of [C II] emission arises from neutral gas without systematic variations. We conclude a coherence exists between the emissions from ionized and neutral gases regarding energy sources and distribution. We argue that [C II] is physically a metallicity-dependent star formation rate (SFR) tracer, while its correlations with atomic or molecular gas masses are secondary. Crucially, the [C II] ``deficit'' is only part of a universal ``deficit'' problem that shows in all neutral and ionized gas lines including extinction-corrected H$α$, caused by infrared (IR) luminosities and characterized by a dichotomy in gas and dust behaviors. This universal ``deficit'' marks a breakdown of the obscuration-corrected star-formation rate (SFR) calibration and imperils SFR estimates. We argue that it is caused by either IR ``excess'' or ionized gas ``deficit'', and present possible scenarios. A renewed picture of ISM structure is needed to reconcile with ionized--neutral gas coherence, metallicity dependence, and gas--dust dichotomy. We also discuss differences of FIR FSL at high redshifts: the offset in ``deficit'' trends, the similar [O III]/[C II] in metal-poor galaxies, and elevated [O III]/[C II] in dusty galaxies.

astro-ph.GA↗

arXiv2025

Fine-structure Line Atlas for Multi-wavelength Extragalactic Study (FLAMES) II: Photoionization Model View of Ionized to Neutral Gas Emission

Far-infrared (FIR) and mid-infrared (MIR) fine-structure lines (FSLs) provide key diagnostics of physical conditions in the interstellar medium (ISM). Building on empirical relations established in our previous work, we use photoionization models to systematically investigate the emission from both ionized and neutral gas phases in a coherent structure. By applying power-law fits to model parameters, we quantitatively capture how key FIR FSL ratios scale with physical properties such as density, radiation field strength and hardness, and elemental abundances. Our analysis confirms the primary dependencies behind most observed empirical trends and establishes certain FIR FSL ratios as tracers of physical parameters, while revealing that parameter marginalization-particularly in density and the O/H-$U$-$Q_1/Q_0$ relation-plays a crucial role in shaping tight correlations seen in galaxies. We also identify persistent challenges, including degeneracies between ionization parameter and radiation field hardness, uncertainties in neutral gas density, and difficulties in modeling dusty H II regions. We outline the fundamental observational and theoretical limitations of current FIR FSL diagnostics, and highlight prospects for advancing the field through comprehensive, multi-wavelength studies of diverse galaxy populations.

astro-ph.GA↗

arXiv2025

Fine-structure Line Atlas for Multi-wavelength Extragalactic Study (FLAMES) I: Comprehensive Low and High Redshift Catalogs and Empirical Relations for Probing Gas Conditions

Far-infrared (FIR) and mid-infrared (MIR) fine-structure lines (FSLs) are widely used for studying galaxies nearby and faraway. However, interpreting these lines is complicated by factors including sample and data bias, mismatch between resolved calibrations and unresolved observations, limitations in generalizing from case studies, and unresolved issues like the origin of [C II] emission and the so-called ``deficit.''In this series of papers, we assemble and analyze the most comprehensive atlas of FSL data to date. We explore their empirical correlations (paper I), compare them with photoionization models that cover multiphase gas (paper II), and discuss their physical origins and the new perspectives they offer for studying physical properties (paper III). The first paper introduces value-added catalogs of global FSL data of low- and high-z galaxies compiled from the literature, covering most of the existing observations, supplemented with ancillary ultraviolet to FIR information. Our analysis focus on commonly used diagnostics, such as electron density, radiation field strength, metallicity, and electron temperature. We present their distributions across different galaxy samples and redshifts, and cross-validate the reliability of these diagnostics in measuring physical conditions. By examining empirical relations, we identify the contribution of active galactic nuclei (AGN) to the FIR FSLs [O III]88 and [O I]63, and reveal a bias in density measurements. FIR FSLs show good concordance with their optical counterparts. Our findings indicate that variations in FSL ratios are primarily driven by the relative abundances of emitting ions, underscoring their value as tracers of metallicity and radiation field strength. Finally, we compare the FIR FSL properties of low- and high-z galaxies, discussing both their similarities and differences.

astro-ph.GA↗

arXiv2025

UniCombine: Unified Multi-Conditional Combination with Diffusion Transformer

With the rapid development of diffusion models in image generation, the demand for more powerful and flexible controllable frameworks is increasing. Although existing methods can guide generation beyond text prompts, the challenge of effectively combining multiple conditional inputs while maintaining consistency with all of them remains unsolved. To address this, we introduce UniCombine, a DiT-based multi-conditional controllable generative framework capable of handling any combination of conditions, including but not limited to text prompts, spatial maps, and subject images. Specifically, we introduce a novel Conditional MMDiT Attention mechanism and incorporate a trainable LoRA module to build both the training-free and training-based versions. Additionally, we propose a new pipeline to construct SubjectSpatial200K, the first dataset designed for multi-conditional generative tasks covering both the subject-driven and spatially-aligned conditions. Extensive experimental results on multi-conditional generation demonstrate the outstanding universality and powerful capability of our approach with state-of-the-art performance.

cs.CV↗

arXiv2025

Coupled Cluster Downfolding Theory in Simulations of Chemical Systems on Quantum Hardware

The practical application of quantum technologies to chemical problems faces significant challenges, particularly in the treatment of realistic basis sets and the accurate inclusion of electron correlation effects. A direct approach to these problems is currently infeasible due to limitations in the number of logical qubits, their fidelity, and the shallow circuit depths supported by existing hardware; all of which hinder simulations at the required level of accuracy. A promising alternative is hybrid quantum-classical computing, where classical resources are used to construct effective Hamiltonians characterized by dimensions that conform to the constraints of current quantum devices. In this paper, we demonstrate the performance of a hybrid approach: coupled-cluster downfolded Hamiltonians are first evaluated in reduced-dimensionality active spaces, and the corresponding ground-state energies are subsequently computed using quantum algorithms. Our comprehensive analysis explores the achievable accuracy in recovering correlation energies when hundreds of orbitals are downfolded into a problem size tractable by today's quantum hardware. We argue that such flexible hybrid algorithms, where problem size can be tailored to available quantum resources, can serve as a bridge between noisy intermediate-scale quantum (NISQ) devices and future fault-tolerant quantum computers, marking a step toward the early realization of quantum advantage in chemistry.

quant-ph↗

arXiv2025

Benchmarking Generalizable Bimanual Manipulation: RoboTwin Dual-Arm Collaboration Challenge at CVPR 2025 MEIS Workshop

Embodied Artificial Intelligence (Embodied AI) is an emerging frontier in robotics, driven by the need for autonomous systems that can perceive, reason, and act in complex physical environments. While single-arm systems have shown strong task performance, collaborative dual-arm systems are essential for handling more intricate tasks involving rigid, deformable, and tactile-sensitive objects. To advance this goal, we launched the RoboTwin Dual-Arm Collaboration Challenge at the 2nd MEIS Workshop, CVPR 2025. Built on the RoboTwin Simulation platform (1.0 and 2.0) and the AgileX COBOT-Magic Robot platform, the competition consisted of three stages: Simulation Round 1, Simulation Round 2, and a final Real-World Round. Participants totally tackled 17 dual-arm manipulation tasks, covering rigid, deformable, and tactile-based scenarios. The challenge attracted 64 global teams and over 400 participants, producing top-performing solutions like SEM and AnchorDP3 and generating valuable insights into generalizable bimanual policy learning. This report outlines the competition setup, task design, evaluation methodology, key findings and future direction, aiming to support future research on robust and generalizable bimanual manipulation policies. The Challenge Webpage is available at https://robotwin-benchmark.github.io/cvpr-2025-challenge/.

cs.RO↗

arXiv2025

AnchorDP3: 3D Affordance Guided Sparse Diffusion Policy for Robotic Manipulation

We present AnchorDP3, a diffusion policy framework for dual-arm robotic manipulation that achieves state-of-the-art performance in highly randomized environments. AnchorDP3 integrates three key innovations: (1) Simulator-Supervised Semantic Segmentation, using rendered ground truth to explicitly segment task-critical objects within the point cloud, which provides strong affordance priors; (2) Task-Conditioned Feature Encoders, lightweight modules processing augmented point clouds per task, enabling efficient multi-task learning through a shared diffusion-based action expert; (3) Affordance-Anchored Keypose Diffusion with Full State Supervision, replacing dense trajectory prediction with sparse, geometrically meaningful action anchors, i.e., keyposes such as pre-grasp pose, grasp pose directly anchored to affordances, drastically simplifying the prediction space; the action expert is forced to predict both robot joint angles and end-effector poses simultaneously, which exploits geometric consistency to accelerate convergence and boost accuracy. Trained on large-scale, procedurally generated simulation data, AnchorDP3 achieves a 98.7% average success rate in the RoboTwin benchmark across diverse tasks under extreme randomization of objects, clutter, table height, lighting, and backgrounds. This framework, when integrated with the RoboTwin real-to-sim pipeline, has the potential to enable fully autonomous generation of deployable visuomotor policies from only scene and instruction, totally eliminating human demonstrations from learning manipulation skills.

cs.RO↗

arXiv2025

Pro-AD: Learning Comprehensive Prototypes with Prototype-based Constraint for Multi-class Unsupervised Anomaly Detection

Prototype-based reconstruction methods for unsupervised anomaly detection utilize a limited set of learnable prototypes which only aggregates insufficient normal information, resulting in undesirable reconstruction. However, increasing the number of prototypes may lead to anomalies being well reconstructed through the attention mechanism, which we refer to as the "Soft Identity Mapping" problem. In this paper, we propose Pro-AD to address these issues and fully utilize the prototypes to boost the performance of anomaly detection. Specifically, we first introduce an expanded set of learnable prototypes to provide sufficient capacity for semantic information. Then we employ a Dynamic Bidirectional Decoder which integrates the process of the normal information aggregation and the target feature reconstruction via prototypes, with the aim of allowing the prototypes to aggregate more comprehensive normal semantic information from different levels of the image features and the target feature reconstruction to not only utilize its contextual information but also dynamically leverage the learned comprehensive prototypes. Additionally, to prevent the anomalies from being well reconstructed using sufficient semantic information through the attention mechanism, Pro-AD introduces a Prototype-based Constraint that applied within the target feature reconstruction process of the decoder, which further improves the performance of our approach. Extensive experiments on multiple challenging benchmarks demonstrate that our Pro-AD achieve state-of-the-art performance, highlighting its superior robustness and practical effectiveness for Multi-class Unsupervised Anomaly Detection task.

cs.CV↗

arXiv2025

Noise-Informed Diffusion-Generated Image Detection with Anomaly Attention

With the rapid development of image generation technologies, especially the advancement of Diffusion Models, the quality of synthesized images has significantly improved, raising concerns among researchers about information security. To mitigate the malicious abuse of diffusion models, diffusion-generated image detection has proven to be an effective countermeasure.However, a key challenge for forgery detection is generalising to diffusion models not seen during training. In this paper, we address this problem by focusing on image noise. We observe that images from different diffusion models share similar noise patterns, distinct from genuine images. Building upon this insight, we introduce a novel Noise-Aware Self-Attention (NASA) module that focuses on noise regions to capture anomalous patterns. To implement a SOTA detection model, we incorporate NASA into Swin Transformer, forming an novel detection architecture NASA-Swin. Additionally, we employ a cross-modality fusion embedding to combine RGB and noise images, along with a channel mask strategy to enhance feature learning from both modalities. Extensive experiments demonstrate the effectiveness of our approach in enhancing detection capabilities for diffusion-generated images. When encountering unseen generation methods, our approach achieves the state-of-the-art performance.Our code is available at https://github.com/WeinanGuan/NASA-Swin.

cs.CV↗

arXiv2025

The FAST Globular Cluster Pulsar Survey (GC FANS)

By January 2025, 60 pulsars were discovered by the Five-hundred-meter Aperture Spherical radio Telescope globular cluster (GC) pulsar survey (GC FANS), with spin periods spanning 1.98 ms to 3960.72 ms. Of these, 55 are millisecond pulsars (MSPs; $P<30$ ms), while 34 are binaries with orbital periods spanning 0.12 days to 466.47 days. This paper describes GC FANS, a deep, thorough search for pulsars in 41 GCs in the FAST sky ($-14^\circ < δ< 65^\circ$) and describes new discoveries in 14 of them. We present updated timing solutions for M92A, NGC 6712A, M71A, and M71E, all of which are ``spider'' pulsars with short orbital periods. We present new timing solutions for M71B, C, and D. With orbital periods of $\sim$466 and 378 days, M71B and M71C are the widest known GC binaries; these systems resemble the normal wide MSP-He WD systems in the Galactic disk. With a spin period of 101 ms, M71D is in an eccentric ($e\sim$0.63) orbit with an 11-day period and a massive companion; the system has a total mass of $2.63 \pm 0.08 \, M_{\odot}$. These features and its large characteristic age suggest it is a double neutron star system (DNS) formed via massive binary evolution early in the cluster's history, akin to Galactic disk DNSs--unlike other candidate GC DNSs, which typically form dynamically. A comparative analysis of GC pulsar populations within FAST's sky reveals that most clusters (10 of 14) resemble the Galactic disk MSP population, likely due to lower stellar densities.

astro-ph.HE↗

arXiv2025

Concept Corrector: Erase concepts on the fly for text-to-image diffusion models

Text-to-image diffusion models have demonstrated the underlying risk of generating various unwanted content, such as sexual elements. To address this issue, the task of concept erasure has been introduced, aiming to erase any undesired concepts that the models can generate. Previous methods, whether training-based or training-free, have primarily focused on the input side, i.e., texts. However, they often suffer from incomplete erasure due to limitations in the generalization from limited prompts to diverse image content. In this paper, motivated by the notion that concept erasure on the output side, i.e., generated images, may be more direct and effective, we propose Concept Corrector. It checks target concepts based on visual features provided by final generated images predicted at certain time steps. Further, it incorporates Concept Removal Attention to erase generated concept features. It overcomes the limitations of existing methods, which are either unable to remove the concept features that have been generated in images or rely on the assumption that the related concept words are contained in input prompts. In the whole pipeline, our method changes no model parameters and only requires a given target concept as well as the corresponding replacement content, which is easy to implement. To the best of our knowledge, this is the first erasure method based on intermediate-generated images, achieving the ability to erase concepts on the fly. The experiments on various concepts demonstrate its impressive erasure performance.

cs.CV↗

arXiv2025

IP-Dialog: Evaluating Implicit Personalization in Dialogue Systems with Synthetic Data

In modern dialogue systems, the ability to implicitly infer user backgrounds from conversations and leverage this information for personalized assistance is crucial. However, the scarcity of high-quality data remains a fundamental challenge to evaluating and improving this capability. Traditional dataset construction methods are labor-intensive, resource-demanding, and raise privacy concerns. To address these issues, we propose a novel approach for automatic synthetic data generation and introduce the Implicit Personalized Dialogue (IP-Dialog) benchmark along with a training dataset, covering 10 tasks and 12 user attribute types. Additionally, we develop a systematic evaluation framework with four metrics to assess both attribute awareness and reasoning capabilities. We further propose five causal graphs to elucidate models' reasoning pathways during implicit personalization. Extensive experiments yield insightful observations and prove the reliability of our dataset.

cs.CL↗

arXiv2025

NTIRE 2025 challenge on Text to Image Generation Model Quality Assessment

This paper reports on the NTIRE 2025 challenge on Text to Image (T2I) generation model quality assessment, which will be held in conjunction with the New Trends in Image Restoration and Enhancement Workshop (NTIRE) at CVPR 2025. The aim of this challenge is to address the fine-grained quality assessment of text-to-image generation models. This challenge evaluates text-to-image models from two aspects: image-text alignment and image structural distortion detection, and is divided into the alignment track and the structural track. The alignment track uses the EvalMuse-40K, which contains around 40K AI-Generated Images (AIGIs) generated by 20 popular generative models. The alignment track has a total of 371 registered participants. A total of 1,883 submissions are received in the development phase, and 507 submissions are received in the test phase. Finally, 12 participating teams submitted their models and fact sheets. The structure track uses the EvalMuse-Structure, which contains 10,000 AI-Generated Images (AIGIs) with corresponding structural distortion mask. A total of 211 participants have registered in the structure track. A total of 1155 submissions are received in the development phase, and 487 submissions are received in the test phase. Finally, 8 participating teams submitted their models and fact sheets. Almost all methods have achieved better results than baseline methods, and the winning methods in both tracks have demonstrated superior prediction performance on T2I model quality assessment.

cs.CV↗