arXiv subjects

Bo Liu

Publications and source records attributed to Bo Liu.

At least 145 records · Page 8Linked to original sources

arXiv2026

IGU-LoRA: Adaptive Rank Allocation via Integrated Gradients and Uncertainty-Aware Scoring

As large language models (LLMs) scale to billions of parameters, full-parameter fine-tuning becomes compute- and memory-prohibitive. Parameter-efficient fine-tuning (PEFT) mitigates this issue by updating only a small set of task-specific parameters while keeping the base model frozen. Among PEFT approaches, low-rank adaptation (LoRA) is widely adopted; however, it enforces a uniform rank across layers despite substantial variation in layer importance, motivating {layerwise} rank allocation. Recent adaptive-rank variants (e.g., AdaLoRA) allocate ranks based on importance scores, yet typically rely on instantaneous gradients that capture only local sensitivity, overlooking non-local, pathwise effects within the same layer, which yields unstable and biased scores. To address this limitation, we introduce IGU-LoRA, an adaptive-rank LoRA that (i) computes within-layer Integrated Gradients (IG) sensitivities and aggregates them into a layer-level score for rank allocation, and (ii) applies an uncertainty-aware scheme using exponential moving averages with deviation tracking to suppress noisy updates and calibrate rank selection. Theoretically, we prove an upper bound on the composite trapezoidal rule approximation error for parameter-space IG under a pathwise Hessian-Lipschitz condition, which informs the quadrature budget. Across diverse tasks and architectures, IGU-LoRA consistently outperforms strong PEFT baselines at matched parameter budgets, improving downstream accuracy and robustness. Ablations confirm the contributions of pathwise within-layer sensitivity estimates and uncertainty-aware selection to effective rank allocation. Our code is publicly available at https://github.com/withyou12/igulora.git

cs.LG↗

arXiv2026

U-Face: An Efficient and Generalizable Framework for Unsupervised Facial Attribute Editing via Subspace Learning

Latent space-based facial attribute editing methods have gained popularity in applications such as digital entertainment, virtual avatar creation, and human-computer interaction systems due to their potential for efficient and flexible attribute manipulation, particularly for continuous edits. Among these, unsupervised latent space-based methods, which discover effective semantic vectors without relying on labeled data, have attracted considerable attention in the research community. However, existing methods still encounter difficulties in disentanglement, as manipulating a specific facial attribute may unintentionally affect other attributes, complicating fine-grained controllability. To address these challenges, we propose a novel framework designed to offer an effective and adaptable solution for unsupervised facial attribute editing, called Unsupervised Facial Attribute Controllable Editing (U-Face). The proposed method frames semantic vector learning as a subspace learning problem, where latent vectors are approximated within a lower-dimensional semantic subspace spanned by a semantic vector matrix. This formulation can also be equivalently interpreted from a projection-reconstruction perspective and further generalized into an autoencoder framework, providing a foundation that can support disentangled representation learning in a flexible manner. To improve disentanglement and controllability, we impose orthogonal non-negative constraints on the semantic vectors and incorporate attribute boundary vectors to reduce entanglement in the learned directions. Although these constraints make the optimization problem challenging, we design an alternating iterative algorithm, called Alternating Iterative Disentanglement and Controllability (AIDC), with closed-form updates and provable convergence under specific conditions.

cs.CV↗

arXiv2026

Examining Reasoning LLMs-as-Judges in Non-Verifiable LLM Post-Training

Reasoning LLMs-as-Judges, which can benefit from inference-time scaling, provide a promising path for extending the success of reasoning models to non-verifiable domains where the output correctness/quality cannot be directly checked. However, while reasoning judges have shown better performance on static evaluation benchmarks, their effectiveness in actual policy training has not been systematically examined. Therefore, we conduct a rigorous study to investigate the actual impact of non-reasoning and reasoning judges in reinforcement-learning-based LLM alignment. Our controlled synthetic setting, where a "gold-standard" judge (gpt-oss-120b) provides preference annotations to train smaller judges, reveals key differences between non-reasoning and reasoning judges: non-reasoning judges lead to reward hacking easily, while reasoning judges can lead to policies that achieve strong performance when evaluated by the gold-standard judge. Interestingly, we find that the reasoning-judge-trained policies achieve such strong performance by learning to generate highly effective adversarial outputs that can also score well on popular benchmarks such as Arena-Hard by deceiving other LLM-judges. Combined with our further analysis, our study highlights both important findings and room for improvements for applying (reasoning) LLM-judges in non-verifiable LLM post-training.

cs.AI↗

arXiv2026

Existence of nonlinearly scalarized black holes in Einstein-scalar-Gauss-Bonnet theory with polynomial couplings

Nonlinearly scalarized black holes are investigated in Einstein-scalar-Gauss-Bonnet (EsGB) theory with polynomial coupling functions $ζ(ϕ)$ satisfying $ζ''(0) = 0$, where $ζ'(ϕ) = 0$ features besides $ϕ=0$ solutions with constant $ϕ_{\rm s} \ne 0$. We determine the threshold amplitudes for Gaussian pulses, above which Schwarzschild black holes (SBHs) %become unstable and may transition to scalarized black holes for two coupling functions: $ζ(ϕ)=αϕ^4-βϕ^8$ and $ζ(ϕ)=αϕ^4-βϕ^6$. In contrast, for the quartic coupling function $ζ(ϕ)=αϕ^4$ SBHs are stable. Treating $ζ(ϕ)R_{GB}^2$ as an effective potential $V_\text{eff}$ provides an explanation for the ``plateau" and the divergence observed in the time evolution. We then construct the branches of nonlinearly scalarized black holes in the probe limit and with backreaction. While the pattern of the solution branches in the probe limit exhibits universal features, the presence of backreaction reveals a distinct dependence on the coupling strength $β$.

gr-qc↗

arXiv2026

MAD: A Multimodal and Multi-perspective Affective Dataset with Hierarchical Annotations

This work presents MAD (Multimodal Affection Dataset), a multimodal emotion dataset designed for affective computing and neurophysiological modeling. MAD is built upon synchronous collection of diverse physiological signals (EEG, ECG, EOG, EMG, PPG, and BCG) together with tri-view RGB-D facial videos, enabling the observation of emotional dynamics from neural, physiological, and behavioral perspectives. The dataset consists of synchronized recordings from 18 participants and introduces two key contributions. First, it provides temporally aligned multimodal data that jointly capture central neural activity, peripheral physiological responses, and overt facial expressions. Second, it incorporates a three-level emotion annotation framework spanning stimulus elicitation, subjective cognition, and behavioral expression, supporting joint modeling of the full emotion process. To validate the dataset, we conduct systematic benchmark experiments covering intra-subject EEG emotion recognition, cross-subject EEG transfer learning, consistency analysis and emotion classification with cardiac-related signals, multimodal physiological fusion, and multi-view facial emotion recognition. The experimental results demonstrate that MAD supports consistent and comparable performance across both unimodal and multimodal settings, establishing it as a reliable benchmark for emotion recognition and cross-modal affective analysis, and as a valuable resource for studying emotion mechanisms across multiple levels.

eess.SP↗

arXiv2026

EasyAnimate: High-Performance Video Generation Framework with Hybrid Windows Attention and Reward Backpropagation

This paper introduces EasyAnimate, an efficient and high quality video generation framework that leverages diffusion transformers to achieve high-quality video production, encompassing data processing, model training, and end-to-end inference. Despite substantial advancements achieved by video diffusion models, existing video generation models still struggles with slow generation speeds and less-than-ideal video quality. To improve training and inference efficiency without compromising performance, we propose Hybrid Window Attention. We design the multidirectional sliding window attention in Hybrid Window Attention, which provides stronger receptive capabilities in 3D dimensions compared to naive one, while reducing the model's computational complexity as the video sequence length increases. To enhance video generation quality, we optimize EasyAnimate using reward backpropagation to better align with human preferences. As a post-training method, it greatly enhances the model's performance while ensuring efficiency. In addition to the aforementioned improvements, EasyAnimate integrates a series of further refinements that significantly improve both computational efficiency and model performance. We introduce a new training strategy called Training with Token Length to resolve uneven GPU utilization in training videos of varying resolutions and lengths, thereby enhancing efficiency. Additionally, we use a multimodal large language model as the text encoder to improve text comprehension of the model. Experiments demonstrate significant enhancements resulting from the above improvements. The EasyAnimate achieves state-of-the-art performance on both the VBench leaderboard and human evaluation. Code and pre-trained models are available at https://github.com/aigc-apps/EasyAnimate.

cs.CV↗

arXiv2026

Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play

Although reinforcement learning (RL) has emerged as a promising approach for improving vision-language models (VLMs) and multimodal large language models (MLLMs), current methods rely heavily on manually curated datasets and costly human verification, which limits scalable self-improvement in multimodal systems. To address this challenge, we propose Vision-Zero, a label-free, domain-agnostic multi-agent self-play framework for self-evolving VLMs through competitive visual games generated from arbitrary image inputs. Specifically, Vision-Zero encompasses three main attributes: (1) Strategic Self-Play Framework: Vision-Zero trains VLMs in "Who Is the Spy"-style games, where the models engage in strategic reasoning and actions across multiple roles. Through interactive gameplay, models autonomously generate their training data without human annotation. (2) Gameplay from Arbitrary Images: Unlike existing gamified frameworks, Vision-Zero can generate games from arbitrary images, thereby enhancing the model's reasoning ability across diverse domains and showing strong generalization to different tasks. We demonstrate this versatility using three distinct types of image datasets: CLEVR-based synthetic scenes, charts, and real-world images. (3) Sustainable Performance Gain: We introduce Iterative Self-Play Policy Optimization (Iterative-SPO), a novel training algorithm that alternates between Self-Play and reinforcement learning with verifiable rewards (RLVR), mitigating the performance plateau often seen in self-play-only training and achieving sustained long-term improvements. Despite using label-free data, Vision-Zero achieves state-of-the-art performance on reasoning, chart question answering, and vision-centric understanding tasks, surpassing other annotation-based methods. Models and code have been released at https://github.com/wangqinsi1/Vision-Zero.

cs.CV↗

arXiv2026

Towards Accurate and Interpretable Time-series Forecasting: A Polynomial Learning Approach

Time series forecasting enables early warning and has driven asset performance management from traditional planned maintenance to predictive maintenance. However, the lack of interpretability in forecasting methods undermines users' trust and complicates debugging for developers. Consequently, interpretable time-series forecasting has attracted increasing research attention. Nevertheless, existing methods suffer from several limitations, including insufficient modeling of temporal dependencies, lack of feature-level interpretability to support early warning, and difficulty in simultaneously achieving the accuracy and interpretability. This paper proposes the interpretable polynomial learning (IPL) method, which integrates interpretability into the model structure by explicitly modeling original features and their interactions of arbitrary order through polynomial representations. This design preserves temporal dependencies, provides feature-level interpretability, and offers a flexible trade-off between prediction accuracy and interpretability by adjusting the polynomial degree. We evaluate IPL on simulated and Bitcoin price data, showing that it achieves high prediction accuracy with superior interpretability compared with widely used explainability methods. Experiments on field-collected antenna data further demonstrate that IPL yields simpler and more efficient early warning mechanisms.

cs.LG↗

arXiv2026

SPIRAL: Self-Play on Zero-Sum Games Incentivizes Reasoning via Multi-Agent Multi-Turn Reinforcement Learning

Recent advances in reinforcement learning have shown that language models can develop sophisticated reasoning through training on tasks with verifiable rewards, but these approaches depend on human-curated problem-answer pairs and domain-specific reward engineering. We introduce SPIRAL, a self-play framework where models learn by playing multi-turn, zero-sum games against continuously improving versions of themselves, generating an automatic curriculum of stronger opponents, and eliminating the need for human supervision. To enable this self-play training at scale, we implement a fully online, multi-turn, multi-agent reinforcement learning system for LLMs and propose role-conditioned advantage estimation (RAE) to stabilize multi-agent training. SPIRAL produces reasoning capabilities that transfer broadly, improving performance by up to 10% across a suite of 8 reasoning benchmarks on 4 different models spanning Qwen and Llama model families, outperforming supervised fine-tuning on 25,000 expert game trajectories. Multi-game training (TicTacToe, Kuhn Poker, Simple Negotiation) yields the strongest results, with improvements observed across both base and instruction-tuned models. Analysis of chain-of-thought traces reveals that games develop distinct cognitive patterns that transfer to improve reasoning performance, with different games developing complementary strengths. Even models which have already been trained on reasoning tasks using RLVR, like DeepSeek-R1-Distill-Qwen-7B, still benefit from our approach. These results demonstrate that zero-sum games naturally develop transferable reasoning capabilities across diverse model architectures and training stages, highlighting a promising direction for autonomous reasoning development. Our code can be found in https://github.com/spiral-rl/spiral.

cs.AI↗

arXiv2026

GEM: A Gym for Agentic LLMs

The training paradigm for large language models (LLMs) is moving from static datasets to experience-based learning, where agents acquire skills via interacting with complex environments. To facilitate this transition we introduce GEM (General Experience Maker), an open-source environment simulator designed for the age of LLMs. Analogous to OpenAI-Gym for traditional reinforcement learning (RL), GEM provides a standardized framework for the environment-agent interface, including asynchronous vectorized execution for high throughput, and flexible wrappers for easy extensibility. GEM also features a diverse suite of environments, robust integrated tools, and single-file example scripts demonstrating using GEM with five popular RL training frameworks. Along with this, we also provide a set of baselines across 24 environments using REINFORCE with Return Batch Normalization (ReBN), which -- unlike GRPO -- is compatible with the full RL setting of dense per-turn rewards and offers better credit assignment. We further conduct apple-to-apple benchmarking of PPO, GRPO and REINFORCE in both single- and multi-turn settings using GEM to shed light on the algorithmic designs. Lastly, GEM also functions as a convenient evaluation toolkit besides a training environment. We hope this framework can help accelerate future agentic LLM research.

cs.LG↗

arXiv2026

All Cities are Equal: A Unified Human Mobility Generation Model Enabled by LLMs

Synthetic human mobility generation is gaining traction as an ethical and practical approach to supporting the data needs of intelligent urban systems. Existing methods perform well primarily in data-rich cities, while their effectiveness declines significantly in cities with limited data resources. However, the ability to generate reliable human mobility data should not depend on a city's size or available resources, all cities deserve equal consideration. To address this open issue, we propose UniMob, a unified human mobility generation model across cities. UniMob is composed of three main components: an LLM-powered travel planner that derives high-level, temporally-aware, and semantically meaningful travel plans; a unified spatial embedding module that projects the spatial regions of various cities into a shared representation space; and a diffusion-based mobility generator that captures the joint spatiotemporal characteristics of human movement, guided by the derived travel plans. We evaluate UniMob extensively using two real-world datasets covering five cities. Comprehensive experiments show that UniMob significantly outperforms state-of-the-art baselines, achieving improvements of over 30\% across multiple evaluation metrics. Further analysis demonstrates UniMob's robustness in both zero- and few-shot scenarios, underlines the importance of LLM guidance, verifies its privacy-preserving nature, and showcases its applicability for downstream tasks.

cs.ET↗

arXiv2026

Co-Propagation of Quantum Time Synchronization and Optical Frequency Transfer over a 122 km Hollow-Core Fiber

The co-propagation of quantum and classical signals through shared optical fibers is crucial for scalable quantum networks. However, this coexistence is fundamentally limited by spontaneous Raman scattering (SpRS) from the bright classical light, which generates overwhelming noise that disrupts the single-photon-level quantum signals. Here, we overcome this long-standing challenge by leveraging the inherently ultralow nonlinearity of hollow-core fiber (HCF) to suppress SpRS noise. By operating both the quantum time synchronization (QTS) and classical optical frequency transfer (OFT) signals within the telecom C-band, separated by only ~10 nm, we successfully demonstrate their simultaneous transmission over a 122-km HCF link. With a classical OFT power of 1 mW, the QTS performance shows negligible degradation, maintaining sub-picosecond time stability at 2000 s, while the OFT achieves a fractional frequency instability of 10^-20. Near-sub-picosecond QTS stability is preserved even when the classical power is increased to 3 mW. Furthermore, simulations based on our experimental data indicate that with next-generation low-loss HCF, the platform can tolerate classical powers beyond 10 mW and extend the QTS range to over 500 km. By realizing a unified quantum-classical time-frequency distribution framework, this work establishes HCF as a highly capable and practical platform for future scalable quantum networks.

quant-ph↗

arXiv2026

OPRIDE: Offline Preference-based Reinforcement Learning via In-Dataset Exploration

Preference-based reinforcement learning (PbRL) can help avoid sophisticated reward designs and align better with human intentions, showing great promise in various real-world applications. However, obtaining human feedback for preferences can be expensive and time-consuming, which forms a strong barrier for PbRL. In this work, we address the problem of low query efficiency in offline PbRL, pinpointing two primary reasons: inefficient exploration and overoptimization of learned reward functions. In response to these challenges, we propose a novel algorithm, \textbf{O}ffline \textbf{P}b\textbf{R}L via \textbf{I}n-\textbf{D}ataset \textbf{E}xploration (OPRIDE), designed to enhance the query efficiency of offline PbRL. OPRIDE consists of two key features: a principled exploration strategy that maximizes the informativeness of the queries and a discount scheduling mechanism aimed at mitigating overoptimization of the learned reward functions. Through empirical evaluations, we demonstrate that OPRIDE significantly outperforms prior methods, achieving strong performance with notably fewer queries. Moreover, we provide theoretical guarantees of the algorithm's efficiency. Experimental results across various locomotion, manipulation, and navigation tasks underscore the efficacy and versatility of our approach.

cs.LG↗

arXiv2026

Cautious Optimizers: Improving Training with One Line of Code

AdamW has been the default optimizer for transformer pretraining. For many years, our community searched for faster and more stable optimizers with only constrained positive outcomes. In this work, we propose a \textbf{one-line modification in Pytorch} to any momentum-based optimizer, which we rename cautious optimizer, e.g. C-AdamW and C-Lion. Our theoretical result shows that this modification preserves Adam's Hamiltonian function and it does not break the convergence guarantee under the Lyapunov analysis. In addition, a whole new family of optimizers is revealed by our theoretical insight. Among them, we pick the simplest one for empirical experiments, showing not only consistent speed-up on LLM pretraining, but also image classification, with minimum extra tuning on hyperparameters. Code is available at https://github.com/kyleliang919/C-Optim.

cs.LG↗

arXiv2026

Comparing Fabrication Workflows in CAD to Support Design Reasoning

When novices fabricate, they start by choosing a workflow (e.g., laser cutting, 3D printing, etc.) and corresponding software from a narrow set they know. As they advance their design, another workflow might better suit their intent, but their models remain committed to the original workflow. This prohibits exploration, which fosters informed decision-making. In this paper, we investigate how CAD interfaces can guide exploration and comparison of workflows. Specifically, comparison can advance users' reasoning about design decisions. We developed a prototype interface, CAMeleon, which lets users compare fabrication workflows. Users load 3D models and preview outcomes from different workflows. We hypothesize that presenting alternative outcomes supports exploration and scaffolds informed decision-making. Upon workflow confirmation, CAMeleon allows users export both machine and human instructions for the chosen fabrication workflow. We interviewed seven fabrication educators to understand how such tools can be integrated into teaching and to demonstrate how we adjust our tool based on their insights. In user evaluation (N = 12), guided comparison helped participants consider a broader range of workflows, reflect on trade-offs, and experiment with new ways of planning.

cs.HC↗

arXiv2026

Demonstration of High-Gain Harmonic Lasing in a Terahertz Free-Electron Laser

Compact Free-Electron Lasers (FELs) offering broad, continuous spectral tunability are traditionally constrained by fixed-parameter magnetic structures and the necessity for high-energy electron beams. High-gain Harmonic Lasing (HL) has long been proposed as a solution to overcome these limitations; however, a robust experimental verification of this principle has remained absent. Here, we report the first experimental demonstration of high-gain HL. By employing a frequency-tunable electron beam density modulation to dominate the fundamental instability, we achieved sustained FEL amplification at the 3rd and 5th harmonics of the wiggler. The HL mode generated output power comparable to conventional fundamental operation with enhanced stability and narrower spectral bandwidth. Notably, we demonstrate that HL extends the spectral coverage by a factor of two under fixed facility constraints, achieving pulse energies up to 540 μJ. These results establish high-gain HL as a versatile mechanism for advancing compact, wavelength-flexible FEL facilities.

physics.acc-ph↗

arXiv2026

Gradient Residual Connections

Existing work has linked properties of a function's gradient to the difficulty of function approximation. Motivated by these insights, we study how gradient information can be leveraged to improve neural network's ability to approximate high-frequency functions, and we propose a gradient-based residual connection as a complement to the standard identity skip connection used in residual networks. We provide simple theoretical intuition for why gradient information can help distinguish inputs and improve the approximation of functions with rapidly varying behaviour. On a synthetic regression task with a high-frequency sinusoidal ground truth, we show that conventional residual connections struggle to capture high-frequency patterns. In contrast, our gradient residual substantially improves approximation quality. We then introduce a convex combination of the standard and gradient residuals, allowing the network to flexibly control how strongly it relies on gradient information. After validating the design choices of our proposed method through an ablation study, we further validate our approach's utility on the single-image super-resolution task, where the underlying function may be high-frequency. Finally, on standard tasks such as image classification and segmentation, our method achieves performance comparable to standard residual networks, suggesting its broad utility.

cs.LG↗

arXiv2026

KidMesh: Computational Mesh Reconstruction for Pediatric Congenital Hydronephrosis Using Deep Neural Networks

Pediatric congenital hydronephrosis (CH) is a common urinary tract disorder, primarily caused by obstruction at the renal pelvis-ureter junction. Magnetic resonance urography (MRU) can visualize hydronephrosis, including renal pelvis and calyces, by utilizing the natural contrast provided by water. Existing voxel-based segmentation approaches can extract CH regions from MRU, facilitating disease diagnosis and prognosis. However, these segmentation methods predominantly focus on morphological features, such as size, shape, and structure. To enable functional assessments, such as urodynamic simulations, external complex post-processing steps are required to convert these results into mesh-level representations. To address this limitation, we propose an end-to-end method based on deep neural networks, namely KidMesh, which could automatically reconstruct CH meshes directly from MRU. Generally, KidMesh extracts feature maps from MRU images and converts them into feature vertices through grid sampling. It then deforms a template mesh according to these feature vertices to generate the specific CH meshes of MRU images. Meanwhile, we develop a novel schema to train KidMesh without relying on accurate mesh-level annotations, which are difficult to obtain due to the sparsely sampled MRU slices. Experimental results show that KidMesh could reconstruct CH meshes in an average of 0.4 seconds, and achieve comparable performance to conventional methods without requiring post-processing. The reconstructed meshes exhibited no self-intersections, with only 3.7% and 0.2% of the vertices having error distances exceeding 3.2mm and 6.4mm, respectively. After rasterization, these meshes achieved a Dice score of 0.86 against manually delineated CH masks. Furthermore, these meshes could be used in renal urine flow simulations, providing valuable urodynamic information for clinical practice.

cs.CV↗