arXiv subjects

Chen Zhang

Publications and source records attributed to Chen Zhang.

At least 325 records · Page 18Linked to original sources

arXiv2024

TS-Align: A Teacher-Student Collaborative Framework for Scalable Iterative Finetuning of Large Language Models

Mainstream approaches to aligning large language models (LLMs) heavily rely on human preference data, particularly when models require periodic updates. The standard process for iterative alignment of LLMs involves collecting new human feedback for each update. However, the data collection process is costly and challenging to scale. To address this issue, we introduce the "TS-Align" framework, which fine-tunes a policy model using pairwise feedback data automatically mined from its outputs. This automatic mining process is efficiently accomplished through the collaboration between a large-scale teacher model and a small-scale student model. The policy fine-tuning process can be iteratively repeated using on-policy generations within our proposed teacher-student collaborative framework. Through extensive experiments, we demonstrate that our final aligned policy outperforms the base policy model with an average win rate of 69.7% across seven conversational or instruction-following datasets. Furthermore, we show that the ranking capability of the teacher is effectively distilled into the student through our pipeline, resulting in a small-scale yet effective reward model for policy model alignment.

cs.CL↗

arXiv2024

Disentangling Age and Identity with a Mutual Information Minimization Approach for Cross-Age Speaker Verification

There has been an increasing research interest in cross-age speaker verification~(CASV). However, existing speaker verification systems perform poorly in CASV due to the great individual differences in voice caused by aging. In this paper, we propose a disentangled representation learning framework for CASV based on mutual information~(MI) minimization. In our method, a backbone model is trained to disentangle the identity- and age-related embeddings from speaker information, and an MI estimator is trained to minimize the correlation between age- and identity-related embeddings via MI minimization, resulting in age-invariant speaker embeddings. Furthermore, by using the age gaps between positive and negative samples, we propose an aging-aware MI minimization loss function that allows the backbone model to focus more on the vocal changes with large age gaps. Experimental results show that the proposed method outperforms other methods on multiple Cross-Age test sets of Vox-CA.

cs.SD↗

arXiv2024

Emo-DPO: Controllable Emotional Speech Synthesis through Direct Preference Optimization

Current emotional text-to-speech (TTS) models predominantly conduct supervised training to learn the conversion from text and desired emotion to its emotional speech, focusing on a single emotion per text-speech pair. These models only learn the correct emotional outputs without fully comprehending other emotion characteristics, which limits their capabilities of capturing the nuances between different emotions. We propose a controllable Emo-DPO approach, which employs direct preference optimization to differentiate subtle emotional nuances between emotions through optimizing towards preferred emotions over less preferred emotional ones. Instead of relying on traditional neural architectures used in existing emotional TTS models, we propose utilizing the emotion-aware LLM-TTS neural architecture to leverage LLMs' in-context learning and instruction-following capabilities. Comprehensive experiments confirm that our proposed method outperforms the existing baselines.

eess.AS↗

arXiv2024

Half-VAE: An Encoder-Free VAE to Bypass Explicit Inverse Mapping

Inference and inverse problems are closely related concepts, both fundamentally involving the deduction of unknown causes or parameters from observed data. Bayesian inference, a powerful class of methods, is often employed to solve a variety of problems, including those related to causal inference. Variational inference, a subset of Bayesian inference, is primarily used to efficiently approximate complex posterior distributions. Variational Autoencoders (VAEs), which combine variational inference with deep learning, have become widely applied across various domains. This study explores the potential of VAEs for solving inverse problems, such as Independent Component Analysis (ICA), without relying on an explicit inverse mapping process. Unlike other VAE-based ICA methods, this approach discards the encoder in the VAE architecture, directly setting the latent variables as trainable parameters. In other words, the latent variables are no longer outputs of the encoder but are instead optimized directly through the objective function to converge to appropriate values. We find that, with a suitable prior setup, the latent variables, represented by trainable parameters, can exhibit mutually independent properties as the parameters converge, all without the need for an encoding process. This approach, referred to as the Half-VAE, bypasses the inverse mapping process by eliminating the encoder. This study demonstrates the feasibility of using the Half-VAE to solve ICA without the need for an explicit inverse mapping process.

stat.ML↗

arXiv2024

A Compressive Memory-based Retrieval Approach for Event Argument Extraction

Recent works have demonstrated the effectiveness of retrieval augmentation in the Event Argument Extraction (EAE) task. However, existing retrieval-based EAE methods have two main limitations: (1) input length constraints and (2) the gap between the retriever and the inference model. These issues limit the diversity and quality of the retrieved information. In this paper, we propose a Compressive Memory-based Retrieval (CMR) mechanism for EAE, which addresses the two limitations mentioned above. Our compressive memory, designed as a dynamic matrix that effectively caches retrieved information and supports continuous updates, overcomes the limitations of the input length. Additionally, after pre-loading all candidate demonstrations into the compressive memory, the model further retrieves and filters relevant information from memory based on the input query, bridging the gap between the retriever and the inference model. Extensive experiments show that our method achieves new state-of-the-art performance on three public datasets (RAMS, WikiEvents, ACE05), significantly outperforming existing retrieval-based EAE methods.

cs.CL↗

arXiv2024

Unveiling the 5$f$ electron hybridization process in UPd$_2$Al$_3$ via ARPES and Time-resolved PES

This study investigates the 5$f$-electron-conduction electron hybridization process in the heavy fermion superconductor UPd$_2$Al$_3$ using a combination of angle-resolved photoemission spectroscopy (ARPES) and time-resolved photoemission spectroscopy (tr-PES). ARPES measurements reveal the formation of a hybridization gap at a temperature of approximately 75 K, which becomes more pronounced as the temperature decreases. Notably, the persistence of a flat U 5$f$ band at temperatures well above the hybridization onset challenges conventional understanding. Our findings demonstrate a non-monotonic temperature dependence of the quasiparticle relaxation time, with an anomalous decrease at 20 K, suggesting complex electronic and magnetic interactions. These findings provide detailed insights into the 5$f$-electron hybridization process in UPd$_2$Al$_3$, with significant implications for the understanding of heavy fermion superconductivity and the role of 5$f$-electron hybridization in uranium-based materials.

cond-mat.str-el↗

arXiv2024

Debate on Graph: a Flexible and Reliable Reasoning Framework for Large Language Models

Large Language Models (LLMs) may suffer from hallucinations in real-world applications due to the lack of relevant knowledge. In contrast, knowledge graphs encompass extensive, multi-relational structures that store a vast array of symbolic facts. Consequently, integrating LLMs with knowledge graphs has been extensively explored, with Knowledge Graph Question Answering (KGQA) serving as a critical touchstone for the integration. This task requires LLMs to answer natural language questions by retrieving relevant triples from knowledge graphs. However, existing methods face two significant challenges: \textit{excessively long reasoning paths distracting from the answer generation}, and \textit{false-positive relations hindering the path refinement}. In this paper, we propose an iterative interactive KGQA framework that leverages the interactive learning capabilities of LLMs to perform reasoning and Debating over Graphs (DoG). Specifically, DoG employs a subgraph-focusing mechanism, allowing LLMs to perform answer trying after each reasoning step, thereby mitigating the impact of lengthy reasoning paths. On the other hand, DoG utilizes a multi-role debate team to gradually simplify complex questions, reducing the influence of false-positive relations. This debate mechanism ensures the reliability of the reasoning process. Experimental results on five public datasets demonstrate the effectiveness and superiority of our architecture. Notably, DoG outperforms the state-of-the-art method ToG by 23.7\% and 9.1\% in accuracy on WebQuestions and GrailQA, respectively. Furthermore, the integration experiments with various LLMs on the mentioned datasets highlight the flexibility of DoG. Code is available at \url{https://github.com/reml-group/DoG}.

cs.CL↗

arXiv2024

The Linked Data Benchmark Council (LDBC): Driving competition and collaboration in the graph data management space

Graph data management is instrumental for several use cases such as recommendation, root cause analysis, financial fraud detection, and enterprise knowledge representation. Efficiently supporting these use cases yields a number of unique requirements, including the need for a concise query language and graph-aware query optimization techniques. The goal of the Linked Data Benchmark Council (LDBC) is to design a set of standard benchmarks that capture representative categories of graph data management problems, making the performance of systems comparable and facilitating competition among vendors. LDBC also conducts research on graph schemas and graph query languages. This paper introduces the LDBC organization and its work over the last decade.

cs.DB↗

arXiv2024

On the cosmological abundance of magnetic monopoles

We demonstrate that Debye shielding cannot be employed to constrain the cosmological abundance of magnetic monopoles, contrary to what is stated in the previous literature. Current model-independent bounds on the monopole abundance are then revisited for unit Dirac magnetic charge. We find that the Andromeda Parker bound can be employed to set an upper limit on the monopole flux at the level of $F_M\lesssim 5.3\times 10^{-19}\,\text{cm}^{-2}\text{s}^{-1}\text{sr}^{-1}$ for a monopole mass $10^{13}\,\text{GeV}/c^2\lesssim m\lesssim 10^{16}\,\text{GeV}/c^2$, which is more stringent than the MACRO direct search limit by two orders of magnitude. This translates into stringent constraints on the monopole density parameter $Ω_M$ at the level of $10^{-7}-10^{-4}$ depending on the mass. For larger monopole masses the scenarios in which magnetic monopoles account for all or the majority of dark matter are disfavored.

hep-ph↗

arXiv2024

MulliVC: Multi-lingual Voice Conversion With Cycle Consistency

Voice conversion aims to modify the source speaker's voice to resemble the target speaker while preserving the original speech content. Despite notable advancements in voice conversion these days, multi-lingual voice conversion (including both monolingual and cross-lingual scenarios) has yet to be extensively studied. It faces two main challenges: 1) the considerable variability in prosody and articulation habits across languages; and 2) the rarity of paired multi-lingual datasets from the same speaker. In this paper, we propose MulliVC, a novel voice conversion system that only converts timbre and keeps original content and source language prosody without multi-lingual paired data. Specifically, each training step of MulliVC contains three substeps: In step one the model is trained with monolingual speech data; then, steps two and three take inspiration from back translation, construct a cyclical process to disentangle the timbre and other information (content, prosody, and other language-related information) in the absence of multi-lingual data from the same speaker. Both objective and subjective results indicate that MulliVC significantly surpasses other methods in both monolingual and cross-lingual contexts, demonstrating the system's efficacy and the viability of the three-step approach with cycle consistency. Audio samples can be found on our demo page (mullivc.github.io).

cs.SD↗

arXiv2024

ICAGC 2024: Inspirational and Convincing Audio Generation Challenge 2024

The Inspirational and Convincing Audio Generation Challenge 2024 (ICAGC 2024) is part of the ISCSLP 2024 Competitions and Challenges track. While current text-to-speech (TTS) technology can generate high-quality audio, its ability to convey complex emotions and controlled detail content remains limited. This constraint leads to a discrepancy between the generated audio and human subjective perception in practical applications like companion robots for children and marketing bots. The core issue lies in the inconsistency between high-quality audio generation and the ultimate human subjective experience. Therefore, this challenge aims to enhance the persuasiveness and acceptability of synthesized audio, focusing on human alignment convincing and inspirational audio generation. A total of 19 teams have registered for the challenge, and the results of the competition and the competition are described in this paper.

eess.AS↗

arXiv2024

Study of Wide-Field-of-View X-ray Observations of the Virgo Cluster Using the Lobster Eye Imager for Astronomy

The Lobster Eye Imager for Astronomy (LEIA) is the pathfinder of the wide-field X-ray telescope used in the Einstein Probe mission. In this study, we present an image of the Virgo Cluster taken by LEIA in the 0.5-4.5 keV band with an exposure time of $\sim$17.3 ks in the central region. This extended emission is generally consistent with the results obtained by ROSAT. However, the field is affected by bright point sources due to the instrument's Point Spread Function (PSF) effect. Through fitting of the LEIA spectrum of the Virgo Cluster, we obtained a temperature of $2.1^{+0.3}_{-0.1}$ keV, which is consistent with the XMM-Newton results ($\sim$2.3 keV). Above 1.6 keV, the spectrum is dominated by the X-ray background. In summary, this study validates LEIA's extended source imaging and spectral resolution capabilities for the first time.

astro-ph.HE↗

arXiv2024

REMAST: Real-time Emotion-based Music Arrangement with Soft Transition

Music as an emotional intervention medium has important applications in scenarios such as music therapy, games, and movies. However, music needs real-time arrangement according to changing emotions, bringing challenges to balance emotion real-time fit and soft emotion transition due to the fine-grained and mutable nature of the target emotion. Existing studies mainly focus on achieving emotion real-time fit, while the issue of smooth transition remains understudied, affecting the overall emotional coherence of the music. In this paper, we propose REMAST to address this trade-off. Specifically, we recognize the last timestep's music emotion and fuse it with the current timestep's input emotion. The fused emotion then guides REMAST to generate the music based on the input melody. To adjust music similarity and emotion real-time fit flexibly, we downsample the original melody and feed it into the generation model. Furthermore, we design four music theory features by domain knowledge to enhance emotion information and employ semi-supervised learning to mitigate the subjective bias introduced by manual dataset annotation. According to the evaluation results, REMAST surpasses the state-of-the-art methods in objective and subjective metrics. These results demonstrate that REMAST achieves real-time fit and smooth transition simultaneously, enhancing the coherence of the generated music.

cs.SD↗

arXiv2024

Functional-Edged Network Modeling

Contrasts with existing works which all consider nodes as functions and use edges to represent the relationships between different functions. We target at network modeling whose edges are functional data and transform the adjacency matrix into a functional adjacency tensor, introducing an additional dimension dedicated to function representation. Tucker functional decomposition is used for the functional adjacency tensor, and to further consider the community between nodes, we regularize the basis matrices to be symmetrical. Furthermore, to deal with irregular observations of the functional edges, we conduct model inference to solve a tensor completion problem. It is optimized by a Riemann conjugate gradient descent method. Besides these, we also derive several theorems to show the desirable properties of the functional edged network model. Finally, we evaluate the efficacy of our proposed model using simulation data and real metro system data from Hong Kong and Singapore.

stat.ML↗

arXiv2024

Constraints of the maximum mass of quark stars based on post-merger evolutions

We semi-analytically investigate the post-merger evolution of the binary quark star merger. The effective-one-body method is employed to estimate the energy and angular momentum dissipation due to gravitational waves in the inspiral phase. Three major mechanisms of energy and angular momentum dissipation are considered in the post-merger phase: mass outflows, neutrinos, and gravitational waves. The proportion of each mechanism could be determined by baryon number, energy and angular momentum conservation laws as well as the equilibrium model for rotating quark stars. Applying this analysis to the GW170817 event suggests two important conclusions: 1) a remnant quark star whose mass is smaller than the maximum mass of a uniformly rotating quark star can collapse before its rotational energy is dissipated via electromagnetic radiation (i.e., $\sim 100\,\mathrm{s}$) as the angular momentum left in the remnant quark star might not be large enough to sustain the additional self-gravity of the supramassive quark star due to the angular momentum dissipation of mass outflows, neutrinos and gravitational waves; 2) considering a general quark star equation of state model, a constraint on the maximum mass of cold and non-rotating quark stars is found as $M_{\mathrm{TOV}}\lesssim2.35^{+0.07}_{-0.17}\,M_{\odot}$, assuming a delayed collapse occurred before a large fraction of the total rotational energy ($\color{blue} \gtrsim 10^{53}\,$erg) of the merger remnant was deposited into the merger environment for the GW170817 event. These constraints could be improved with future merger events, once there are more evidences on its post-merger evolution channel or information on the amount of post-merger gravitational wave and neutrino emissions inferred from the multi-messenger observations.

astro-ph.HE↗

arXiv2024

On the energy budget of starquake-induced repeating fast radio bursts

With a growing sample of fast radio bursts (FRBs), we investigate the energy budget of different power sources within the framework of magnetar starquake triggering mechanism. During a starquake, the energy can be released in any form through strain, magnetic, rotational, and gravitational energies. The strain energy can be converted from other three kinds of energy during starquakes. The following findings are revealed: 1. The crust can store free magnetic energy of $\sim10^{46}$ erg by existing toroidal fields, sustaining $10^6$ bursts with frequent starquakes occurring due to crustal instability. 2. The strain energy develops as a rigid object spins down, which can be released during a global starquake accompanied by a glitch. However, it takes a long time to accumulate enough strain energy via spindown. 3. The rotational energy of a magnetar with $P\lesssim0.1\rm\,s$ can match the energy and luminosity budget of FRBs. 4. The budget of the total gravitational energy is high, but the mechanism and efficiency of converting this energy to radiation deserve further exploration.

astro-ph.HE↗

arXiv2024

Infinite-LLM: Efficient LLM Service for Long Context with DistAttention and Distributed KVCache

Large Language Models (LLMs) demonstrate substantial potential across a diverse array of domains via request serving. However, as trends continue to push for expanding context sizes, the autoregressive nature of LLMs results in highly dynamic behavior of the attention layers, showcasing significant differences in computational characteristics and memory requirements from the non-attention layers. This presents substantial challenges for resource management and performance optimization in service systems. Existing static model parallelism and resource allocation strategies fall short when dealing with this dynamicity. To address the issue, we propose Infinite-LLM, a novel LLM serving system designed to effectively handle dynamic context lengths. Infinite-LLM disaggregates attention layers from an LLM's inference process, facilitating flexible and independent resource scheduling that optimizes computational performance and enhances memory utilization jointly. By leveraging a pooled GPU memory strategy across a cluster, Infinite-LLM not only significantly boosts system throughput but also supports extensive context lengths. Evaluated on a dataset with context lengths ranging from a few to 2000K tokens across a cluster with 32 A100 GPUs, Infinite-LLM demonstrates throughput improvement of 1.35-3.4x compared to state-of-the-art methods, enabling efficient and elastic LLM deployment.

cs.DC↗

arXiv2024

DynaThink: Fast or Slow? A Dynamic Decision-Making Framework for Large Language Models

Large language models (LLMs) have demonstrated emergent capabilities across diverse reasoning tasks via popular Chains-of-Thought (COT) prompting. However, such a simple and fast COT approach often encounters limitations in dealing with complicated problems, while a thorough method, which considers multiple reasoning pathways and verifies each step carefully, results in slower inference. This paper addresses the challenge of enabling LLMs to autonomously select between fast and slow inference methods, thereby optimizing both efficiency and effectiveness. We introduce a dynamic decision-making framework that categorizes tasks into two distinct pathways: 'Fast', designated for tasks where the LLM quickly identifies a high-confidence solution, and 'Slow', allocated for tasks that the LLM perceives as complex and for which it has low confidence in immediate solutions as well as requiring more reasoning paths to verify. Experiments on five popular reasoning benchmarks demonstrated the superiority of the DynaThink over baselines.

cs.CL↗