arXiv subjects

Liang Chen

Publications and source records attributed to Liang Chen.

At least 217 records · Page 12Linked to original sources

arXiv2025

Proto-OOD: Enhancing OOD Object Detection with Prototype Feature Similarity

Neural networks that are trained on limited category samples often mispredict out-of-distribution (OOD) objects. We observe that features of the same category are more tightly clustered in feature space, while those of different categories are more dispersed. Based on this, we propose using prototype similarity for OOD detection. Drawing on widely used prototype features in few-shot learning, we introduce a novel OOD detection network structure (Proto-OOD). Proto-OOD enhances the representativeness of category prototypes using contrastive loss and detects OOD data by evaluating the similarity between input features and category prototypes. During training, Proto-OOD generates OOD samples for training the similarity module with a negative embedding generator. When Pascal VOC are used as the in-distribution dataset and MS-COCO as the OOD dataset, Proto-OOD significantly reduces the FPR (false positive rate). Moreover, considering the limitations of existing evaluation metrics, we propose a more reasonable evaluation protocol. The code will be released.

cs.CV↗

arXiv2025

Aubin Property and Strong Regularity Are Equivalent for Nonlinear Second-Order Cone Programming

This paper solves a fundamental open problem in variational analysis on the equivalence between the Aubin property and the strong regularity for nonlinear second-order cone programming (SOCP) at a locally optimal solution. We achieve this by introducing a reduction approach to the Aubin property characterized by the Mordukhovich criterion and a lemma of alternative choices on cones to replace the S-lemma used in Outrata and Ramírez [SIAM J. Optim. 21 (2011) 789-823] and Opazo, Outrata, and Ramírez [SIAM J. Optim. 27 (2017) 2141-2151], where the same SOCP was considered under the strict complementarity condition except for possibly only one block of constraints. As a byproduct, we also offer a new approach to the well-known result of Dontchev and Rockafellar [SIAM J. Optim. 6 (1996) 1087-1105] on the equivalence of the two concepts in conventional nonlinear programming.

math.OC↗

arXiv2025

Characterizations of the Aubin Property of the Solution Mapping for Nonlinear Semidefinite Programming

In this paper, we study the Aubin property of the Karush-Kuhn-Tucker solution mapping for the nonlinear semidefinite programming (NLSDP) problem at a locally optimal solution. In the literature, it is known that the Aubin property implies the constraint nondegeneracy by Fusek [SIAM J. Optim. 23 (2013), pp. 1041-1061] and the second-order sufficient condition by Ding et al. [SIAM J. Optim. 27 (2017), pp. 67-90]. Based on the Mordukhovich criterion, here we further prove that the strong second-order sufficient condition is also necessary for the Aubin property to hold. Consequently, several equivalent conditions including the strong regularity are established for NLSDP's Aubin property. Together with the recent progress made by Chen et al. on the equivalence between the Aubin property and the strong regularity for nonlinear second-order cone programming [SIAM J. Optim., in press; arXiv:2406.13798v3 (2024)], this paper constitutes a significant step forward in characterizing the Aubin property for general non-polyhedral $C^2$-cone reducible constrained optimization problems.

math.OC↗

arXiv2025

Group Sparse-based Tensor CP Decomposition: Model, Algorithms, and Applications in Chemometrics

The CANDECOMP/PARAFAC (or Canonical polyadic, CP) decomposition of tensors has numerous applications in various fields, such as chemometrics, signal processing, machine learning, etc. Tensor CP decomposition assumes the knowledge of the exact CP rank, i.e., the total number of rank-one components of a tensor. However, accurately estimating the CP rank is very challenging. In this work, to address this issue, we prove that the CP rank can be exactly estimated by minimizing the group sparsity of any one of the factor matrices under the unit length constraints on the columns of the other factor matrices. Based on this result, we propose a CP decomposition model with group sparse regularization, which integrates the rank estimation and the tensor decomposition as an optimization problem, whose set of optimal solutions is proved to be nonempty. To solve the proposed model, we propose a double-loop block-coordinate proximal gradient descent algorithm with extrapolation and prove that each accumulation point of the sequence generated by the algorithm is a stationary point of the proposed model. Furthermore, we incorporate a rank reduction strategy into the algorithm to reduce the computational complexity. Finally, we apply the proposed model and algorithms to the component separation problem in chemometrics using real data. Numerical experiments demonstrate the robustness and effectiveness of the proposed methods.

math.NA↗

arXiv2024

Next Token Prediction Towards Multimodal Intelligence: A Comprehensive Survey

Building on the foundations of language modeling in natural language processing, Next Token Prediction (NTP) has evolved into a versatile training objective for machine learning tasks across various modalities, achieving considerable success. As Large Language Models (LLMs) have advanced to unify understanding and generation tasks within the textual modality, recent research has shown that tasks from different modalities can also be effectively encapsulated within the NTP framework, transforming the multimodal information into tokens and predict the next one given the context. This survey introduces a comprehensive taxonomy that unifies both understanding and generation within multimodal learning through the lens of NTP. The proposed taxonomy covers five key aspects: Multimodal tokenization, MMNTP model architectures, unified task representation, datasets \& evaluation, and open challenges. This new taxonomy aims to aid researchers in their exploration of multimodal intelligence. An associated GitHub repository collecting the latest papers and repos is available at https://github.com/LMM101/Awesome-Multimodal-Next-Token-Prediction

cs.CL↗

arXiv2024

Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models

Recent advancements in large language models (LLMs) have led to significant breakthroughs in mathematical reasoning capabilities. However, existing benchmarks like GSM8K or MATH are now being solved with high accuracy (e.g., OpenAI o1 achieves 94.8\% on MATH dataset), indicating their inadequacy for truly challenging these models. To bridge this gap, we propose a comprehensive and challenging benchmark specifically designed to assess LLMs' mathematical reasoning at the Olympiad level. Unlike existing Olympiad-related benchmarks, our dataset focuses exclusively on mathematics and comprises a vast collection of 4428 competition-level problems with rigorous human annotation. These problems are meticulously categorized into over 33 sub-domains and span more than 10 distinct difficulty levels, enabling a holistic assessment of model performance in Olympiad-mathematical reasoning. Furthermore, we conducted an in-depth analysis based on this benchmark. Our experimental results show that even the most advanced models, OpenAI o1-mini and OpenAI o1-preview, struggle with highly challenging Olympiad-level problems, with 60.54\% and 52.55\% accuracy, highlighting significant challenges in Olympiad-level mathematical reasoning.

cs.CL↗

arXiv2024

SCKF-LSTM Based Trajectory Tracking for Electricity-Gas Integrated Energy System

This paper introduces a novel approach for tracking the dynamic trajectories of integrated natural gas and power systems, leveraging a Kalman filter-based structure. To predict the states of the system, the Holt's exponential smoothing techniques and nonlinear dynamic equations of gas pipelines are applied to establish the power and gas system equations, respectively. The square-root cubature Kalman filter algorithm is utilized to address the numerical challenges posed by the strongly nonlinear system equations. The boundary conditions in the gas system include the flow balances at sink nodes, and the mass flow rates of loads have to be predicted at each computation step. For the prediction of load mass flows, the long short-term memory network is employed, known for its effectiveness in time series prediction. Consequently, a combined method based on the square-root cubature Kalman filter and the long short-term memory network is proposed for tracking integrated gas and power systems. To evaluate the tracking performances of the proposed method, the IEEE-39 bus power system and GasLib-40 node gas system are used to form the testing system. Simulation results demonstrate high precision in tracking the dynamic states of power and gas systems. Two indexes are introduced for a numerical analysis of the tracking results, indicating that the accuracy of this method surpasses that of traditional measurements.

eess.SY↗

arXiv2024

UltraEdit: Instruction-based Fine-Grained Image Editing at Scale

This paper presents UltraEdit, a large-scale (approximately 4 million editing samples), automatically generated dataset for instruction-based image editing. Our key idea is to address the drawbacks in existing image editing datasets like InstructPix2Pix and MagicBrush, and provide a systematic approach to producing massive and high-quality image editing samples. UltraEdit offers several distinct advantages: 1) It features a broader range of editing instructions by leveraging the creativity of large language models (LLMs) alongside in-context editing examples from human raters; 2) Its data sources are based on real images, including photographs and artworks, which provide greater diversity and reduced bias compared to datasets solely generated by text-to-image models; 3) It also supports region-based editing, enhanced by high-quality, automatically produced region annotations. Our experiments show that canonical diffusion-based editing baselines trained on UltraEdit set new records on MagicBrush and Emu-Edit benchmarks. Our analysis further confirms the crucial role of real image anchors and region-based editing data. The dataset, code, and models can be found in https://ultra-editing.github.io.

cs.CV↗

arXiv2024

Spectral Energy Distribution Variability of the Blazar OJ 287 during 2009-2021

Using nearly simultaneous radio, near-infrared, optical, and ultraviolet data collected since 2009, we constructed 106 spectral energy distributions (SEDs) of the blazar OJ 287. These SEDs were well-fitted by a log-parabolic model. By classifying the data into `flare' and `quiescent' segments, we found that the median flux at peak frequency of the SEDs during flare segments was 0.37$\pm$0.22 dex higher compared to quiescent segments, while no significant differences were observed in the median values of the curvature parameter $b$ or the peak frequency $\log ν_{\mathrm{p}}$. A significant bluer-when-brighter trend was confirmed through a relation between $V$ magnitude and $B-V$ color index, with this trend being stronger in the flare segments. Additionally, a significant anti-correlation was detected between $\log ν_{\mathrm{p}}$ and $b$, with a slope of 5.79 in the relation between $1/b$ and $\log ν_{\mathrm{p}}$, closer to the prediction from a statistical acceleration model other than a stochastic acceleration interpretation, though a notable discrepancy persists. This discrepancy indicates that additional factors, such as deviations from idealized conditions or radiative contributions-such as thermal emission from the accretion disk in the optical-UV range during quiescent states-may play a role in producing the observed steeper slope. Within the framework of statistical acceleration mechanism, lack of correlation between change in peak intensity and change in peak frequency suggests that change in electron energy distribution is unlikely to be responsible for the time-dependent SED changes. Instead, changes in Doppler boosting or magnetic fields may have a greater influence.

astro-ph.HE↗

arXiv2024

Two-dimensional \b{eta}-phase copper iodide: a promising candidate for low-temperature thermoelectric applications

Bismuth telluride-based materials is the only commercially viable room-temperature thermoelectric material, despite its limited tellurium and poor mechanical properties. The search for materials with a high figure of merit (zT > 1.00) near room temperature remains a major challenge. In this work, we systematically investigate the structural stability and the thermoelectric capabilities of monolayer \b{eta}-CuI and γ-CuI through the density functional theory (DFT) combined with Boltzmann transport theory. Based on the thermoelectric transport coefficients of monolayer \b{eta}-CuI and γ-CuI, we predict their zT values will vary with carrier concentration and increase with temperature. Comparing the zT values, monolayer \b{eta}-CuI demonstrates superior thermoelectric properties compared to γ-CuI. At room temperature, the optimal zT values of monolayer \b{eta}-CuI exceed 1.50, with particularly high values of 2.98 (p-type) and 4.10 (n-type) along the Zigzag direction, demonstrating significant anisotropy. These results suggest the great potential of the monolayer \b{eta}-CuI is promising candidate materials for low temperature thermoelectric applications.

cond-mat.mtrl-sci↗

arXiv2024

LHAASO detection of very-high-energy gamma-ray emission surrounding PSR J0248+6021

We report the detection of an extended very-high-energy (VHE) gamma-ray source coincident with the location of middle-aged (62.4~\rm kyr) pulsar PSR J0248+6021, by using the LHAASO-WCDA data of live 796 days and LHAASO-KM2A data of live 1216 days. A significant excess of \gray induced showers is observed both by WCDA in energy bands of 1-25~\rm TeV and KM2A in energy bands of $>$ 25~\rm TeV with 7.3 $σ$ and 13.5 $σ$, respectively. The best-fit position derived through WCDA data is R.A. = 42.06$^\circ \pm$ 0.12$^\circ$ and Dec. = 60.24$^\circ \pm $ 0.13$^\circ$ with an extension of 0.69$^\circ\pm$0.15$^\circ$ and that of the KM2A data is R.A.= 42.29$^\circ \pm $ 0.13$^\circ$ and Dec. = 60.38$^\circ \pm$ 0.07$^\circ$ with an extension of 0.37$^\circ\pm$0.07$^\circ$. No clear extended multiwavelength counterpart of this LHAASO source has been found from the radio band to the GeV band. The most plausible explanation of the VHE \gray emission is the inverse Compton process of highly relativistic electrons and positrons injected by the pulsar. These electrons/positrons are hypothesized to be either confined within the pulsar wind nebula or to have already escaped into the interstellar medium, forming a pulsar halo.

astro-ph.HE↗

arXiv2024

Multifunctional metalens for trapping and characterizing single atoms

Precise control and manipulation of neutral atoms are essential for quantum technologies but largely dependent on conventional bulky optical setups. Here, we demonstrate a multifunctional metalens that integrates an achromatic lens with large numerical aperture, a quarter-wave plate, and a polarizer for trapping and characterizing single Rubidium atoms. The metalens simultaneously focuses a trapping beam at 852\,nm and collects single-photon fluorescence at 780\,nm. We observe a strong dependence of the trapping lifetime on an external bias magnetic field, suggests a complex interplay between the circularly polarized trapping light and the atom's internal states. Our work showcases the potential of metasurfaces in realizing compact and integrated quantum systems based on cold atoms, opening up new possibilities for studying quantum control and manipulation at the nanoscale.

quant-ph↗

arXiv2024

Towards a Unified View of Preference Learning for Large Language Models: A Survey

Large Language Models (LLMs) exhibit remarkably powerful capabilities. One of the crucial factors to achieve success is aligning the LLM's output with human preferences. This alignment process often requires only a small amount of data to efficiently enhance the LLM's performance. While effective, research in this area spans multiple domains, and the methods involved are relatively complex to understand. The relationships between different methods have been under-explored, limiting the development of the preference alignment. In light of this, we break down the existing popular alignment strategies into different components and provide a unified framework to study the current alignment strategies, thereby establishing connections among them. In this survey, we decompose all the strategies in preference learning into four components: model, data, feedback, and algorithm. This unified view offers an in-depth understanding of existing alignment algorithms and also opens up possibilities to synergize the strengths of different strategies. Furthermore, we present detailed working examples of prevalent existing algorithms to facilitate a comprehensive understanding for the readers. Finally, based on our unified perspective, we explore the challenges and future research directions for aligning large language models with human preferences.

cs.CL↗

arXiv2024

State Space Models on Temporal Graphs: A First-Principles Study

Over the past few years, research on deep graph learning has shifted from static graphs to temporal graphs in response to real-world complex systems that exhibit dynamic behaviors. In practice, temporal graphs are formalized as an ordered sequence of static graph snapshots observed at discrete time points. Sequence models such as RNNs or Transformers have long been the predominant backbone networks for modeling such temporal graphs. Yet, despite the promising results, RNNs struggle with long-range dependencies, while transformers are burdened by quadratic computational complexity. Recently, state space models (SSMs), which are framed as discretized representations of an underlying continuous-time linear dynamical system, have garnered substantial attention and achieved breakthrough advancements in independent sequence modeling. In this work, we undertake a principled investigation that extends SSM theory to temporal graphs by integrating structural information into the online approximation objective via the adoption of a Laplacian regularization term. The emergent continuous-time system introduces novel algorithmic challenges, thereby necessitating our development of GraphSSM, a graph state space model for modeling the dynamics of temporal graphs. Extensive experimental results demonstrate the effectiveness of our GraphSSM framework across various temporal graph benchmarks.

cs.LG↗

arXiv2024

WildOcc: A Benchmark for Off-Road 3D Semantic Occupancy Prediction

3D semantic occupancy prediction is an essential part of autonomous driving, focusing on capturing the geometric details of scenes. Off-road environments are rich in geometric information, therefore it is suitable for 3D semantic occupancy prediction tasks to reconstruct such scenes. However, most of researches concentrate on on-road environments, and few methods are designed for off-road 3D semantic occupancy prediction due to the lack of relevant datasets and benchmarks. In response to this gap, we introduce WildOcc, to our knowledge, the first benchmark to provide dense occupancy annotations for off-road 3D semantic occupancy prediction tasks. A ground truth generation pipeline is proposed in this paper, which employs a coarse-to-fine reconstruction to achieve a more realistic result. Moreover, we introduce a multi-modal 3D semantic occupancy prediction framework, which fuses spatio-temporal information from multi-frame images and point clouds at voxel level. In addition, a cross-modality distillation function is introduced, which transfers geometric knowledge from point clouds to image features.

cs.CV↗

arXiv2024

LFME: A Simple Framework for Learning from Multiple Experts in Domain Generalization

Domain generalization (DG) methods aim to maintain good performance in an unseen target domain by using training data from multiple source domains. While success on certain occasions are observed, enhancing the baseline across most scenarios remains challenging. This work introduces a simple yet effective framework, dubbed learning from multiple experts (LFME), that aims to make the target model an expert in all source domains to improve DG. Specifically, besides learning the target model used in inference, LFME will also train multiple experts specialized in different domains, whose output probabilities provide professional guidance by simply regularizing the logit of the target model. Delving deep into the framework, we reveal that the introduced logit regularization term implicitly provides effects of enabling the target model to harness more information, and mining hard samples from the experts during training. Extensive experiments on benchmarks from different DG tasks demonstrate that LFME is consistently beneficial to the baseline and can achieve comparable performance to existing arts. Code is available at~\url{https://github.com/liangchen527/LFME}.

cs.LG↗

arXiv2024

VLFeedback: A Large-Scale AI Feedback Dataset for Large Vision-Language Models Alignment

As large vision-language models (LVLMs) evolve rapidly, the demand for high-quality and diverse data to align these models becomes increasingly crucial. However, the creation of such data with human supervision proves costly and time-intensive. In this paper, we investigate the efficacy of AI feedback to scale supervision for aligning LVLMs. We introduce VLFeedback, the first large-scale vision-language feedback dataset, comprising over 82K multi-modal instructions and comprehensive rationales generated by off-the-shelf models without human annotations. To evaluate the effectiveness of AI feedback for vision-language alignment, we train Silkie, an LVLM fine-tuned via direct preference optimization on VLFeedback. Silkie showcases exceptional performance regarding helpfulness, visual faithfulness, and safety metrics. It outperforms its base model by 6.9\% and 9.5\% in perception and cognition tasks, reduces hallucination issues on MMHal-Bench, and exhibits enhanced resilience against red-teaming attacks. Furthermore, our analysis underscores the advantage of AI feedback, particularly in fostering preference diversity to deliver more comprehensive improvements. Our dataset, training code and models are available at https://vlf-silkie.github.io.

cs.CV↗

arXiv2024

Athena: Retrieval-augmented Legal Judgment Prediction with Large Language Models

Recently, large language models (LLMs) like ChatGPT, LLaMA, and Claude have prevailed in countless domains, including legal scenarios. With LLMs' rapid technological progress, the development of prompt engineering (PE) as an interface between the LLMs and real-world applications has drawn the attention of all developers. Various PE methods have been proposed to overcome real-world challenges, such as few-shot prompting, chain-of-thought, and retrieval-augmented generation (RAG). However, RAG for legal judgment prediction (LJP) is still underexplored. To address this, we propose "Athena", a novel framework cultivating RAG as a core preprocess component to enhance LLMs' performance on specialized tasks. Athena constructs a knowledge base for accusations, attached with a semantic retrieval mechanism through vectorization. Our experiments show that Athena's overall performance has improved significantly, achieving state-of-the-art results on the CAIL2018 dataset. Our ablation study on the in-context window size parameter further reproduces LLMs' "lost-in-the-middle" phenomenon with a relative positional variation. And with moderate hyper-parameter-tuning, we can achieve at most 95% of accuracy accordingly. We also study the impact of query rewriting and data distribution, providing possible directions for future research based on former analyses.

cs.CL↗