arXiv subjects

Yan Liu

Publications and source records attributed to Yan Liu.

At least 271 records · Page 15Linked to original sources

arXiv2023

Secrets of RLHF in Large Language Models Part I: PPO

Large language models (LLMs) have formulated a blueprint for the advancement of artificial general intelligence. Its primary objective is to function as a human-centric (helpful, honest, and harmless) assistant. Alignment with humans assumes paramount significance, and reinforcement learning with human feedback (RLHF) emerges as the pivotal technological paradigm underpinning this pursuit. Current technical routes usually include \textbf{reward models} to measure human preferences, \textbf{Proximal Policy Optimization} (PPO) to optimize policy model outputs, and \textbf{process supervision} to improve step-by-step reasoning capabilities. However, due to the challenges of reward design, environment interaction, and agent training, coupled with huge trial and error cost of large language models, there is a significant barrier for AI researchers to motivate the development of technical alignment and safe landing of LLMs. The stable training of RLHF has still been a puzzle. In the first report, we dissect the framework of RLHF, re-evaluate the inner workings of PPO, and explore how the parts comprising PPO algorithms impact policy agent training. We identify policy constraints being the key factor for the effective implementation of the PPO algorithm. Therefore, we explore the PPO-max, an advanced version of PPO algorithm, to efficiently improve the training stability of the policy model. Based on our main results, we perform a comprehensive analysis of RLHF abilities compared with SFT models and ChatGPT. The absence of open-source implementations has posed significant challenges to the investigation of LLMs alignment. Therefore, we are eager to release technical reports, reward models and PPO codes, aiming to make modest contributions to the advancement of LLMs.

cs.CL↗

arXiv2023

SAITS: Self-Attention-based Imputation for Time Series

Missing data in time series is a pervasive problem that puts obstacles in the way of advanced analysis. A popular solution is imputation, where the fundamental challenge is to determine what values should be filled in. This paper proposes SAITS, a novel method based on the self-attention mechanism for missing value imputation in multivariate time series. Trained by a joint-optimization approach, SAITS learns missing values from a weighted combination of two diagonally-masked self-attention (DMSA) blocks. DMSA explicitly captures both the temporal dependencies and feature correlations between time steps, which improves imputation accuracy and training speed. Meanwhile, the weighted-combination design enables SAITS to dynamically assign weights to the learned representations from two DMSA blocks according to the attention map and the missingness information. Extensive experiments quantitatively and qualitatively demonstrate that SAITS outperforms the state-of-the-art methods on the time-series imputation task efficiently and reveal SAITS' potential to improve the learning performance of pattern recognition models on incomplete time-series data from the real world. The code is open source on GitHub at https://github.com/WenjieDu/SAITS.

cs.LG↗

arXiv2023

DSRM: Boost Textual Adversarial Training with Distribution Shift Risk Minimization

Adversarial training is one of the best-performing methods in improving the robustness of deep language models. However, robust models come at the cost of high time consumption, as they require multi-step gradient ascents or word substitutions to obtain adversarial samples. In addition, these generated samples are deficient in grammatical quality and semantic consistency, which impairs the effectiveness of adversarial training. To address these problems, we introduce a novel, effective procedure for instead adversarial training with only clean data. Our procedure, distribution shift risk minimization (DSRM), estimates the adversarial loss by perturbing the input data's probability distribution rather than their embeddings. This formulation results in a robust model that minimizes the expected global loss under adversarial attacks. Our approach requires zero adversarial samples for training and reduces time consumption by up to 70\% compared to current best-performing adversarial training methods. Experiments demonstrate that DSRM considerably improves BERT's resistance to textual adversarial attacks and achieves state-of-the-art robust accuracy on various benchmarks.

cs.CL↗

arXiv2023

Engineering Perovskite Emissions via Optical Quasi-Bound-States-in-the-Continuum

Metal halide perovskite quantum dots (PQDs) have emerged as promising materials due to their exceptional photoluminescence (PL) properties. A wide range of applications could benefit from adjustable luminescence properties, while preserving the physical and chemical properties of the PQDs. Therefore, post-synthesis engineering has gained attention recently, involving the use of ion-exchange or external stimuli, such as extreme pressure, magnetic and electric fields. Nevertheless, these methods typically suffer from spectrum broadening, intensity quenching or yield multiple bands. Alternatively, photonic antennas can modify the radiative decay channel of perovskites via the Purcell effect, with the largest wavelength shift being 8 nm to date, at an expense of 5-fold intensity loss. Here, we present an optical nanoantenna array with polarization-controlled quasi-bound-states-in-the-continuum (q-BIC) resonances, which can engineer and shift the photoluminescence wavelength over a ~39 nm range and confers a 21-fold emission enhancement of FAPbI3 perovskite QDs. The spectrum is engineered in a non-invasive manner via lithographically defined antennas and the pump laser polarization at ambient conditions. Our research provides a path towards advanced optoelectronic devices, such as spectrally tailored quantum emitters and lasers.

physics.optics↗

arXiv2023

On the Generalization and Advancement of Half-Sine-Based Pulse Shaping Filters for Constant Envelope OQPSK Modulation

The offset quadrature phase-shift keying (OQPSK) modulation is a key factor for the technique of ZigBee, which has been adopted in IEEE 802.15.4 for wireless communications of Internet of Things (IoT) and Internet of Vehicles (IoV), etc. In this paper, we propose the general conditions of pulse shaping filters (PSFs) with constant envelope (CE) property for OQPSK modulation, which can be easily leveraged to design the PSFs with CE property. Based on these conditions, we further design an advanced PSF called $α$-half-sine PSF. It is verified that the newly designed $α$-half-sine PSF can not only keep the CE property for OQPSK but also achieve better performance than the traditional PSFs in certain scenarios. Moreover, the $α$-half-sine PSF can be simply adjusted to achieve a flexible performance tradeoff between the transition roll-off speed and out-of-band leakage.

eess.SP↗

arXiv2023

Uncovering and Categorizing Social Biases in Text-to-SQL

Content Warning: This work contains examples that potentially implicate stereotypes, associations, and other harms that could be offensive to individuals in certain social groups.} Large pre-trained language models are acknowledged to carry social biases towards different demographics, which can further amplify existing stereotypes in our society and cause even more harm. Text-to-SQL is an important task, models of which are mainly adopted by administrative industries, where unfair decisions may lead to catastrophic consequences. However, existing Text-to-SQL models are trained on clean, neutral datasets, such as Spider and WikiSQL. This, to some extent, cover up social bias in models under ideal conditions, which nevertheless may emerge in real application scenarios. In this work, we aim to uncover and categorize social biases in Text-to-SQL models. We summarize the categories of social biases that may occur in structured data for Text-to-SQL models. We build test benchmarks and reveal that models with similar task accuracy can contain social biases at very different rates. We show how to take advantage of our methodology to uncover and assess social biases in the downstream Text-to-SQL task. We will release our code and data.

cs.CL↗

arXiv2023

Integer k-matching preclusion of graphs

As a generalization of matching preclusion number of a graph, we provide the (strong) integer $k$-matching preclusion number, abbreviated as $MP^{k}$ number ($SMP^{k}$ number), which is the minimum number of edges (vertices and edges) whose deletion results in a graph that has neither perfect integer $k$-matching nor almost perfect integer $k$-matching. In this paper, we show that when $k$ is even, the ($SMP^{k}$) $MP^{k}$ number is equal to the (strong) fractional matching preclusion number. We obtain a necessary condition of graphs with an almost-perfect integer $k$-matching and a relational expression between the matching number and the integer $k$-matching number of bipartite graphs. Thus the $MP^{k}$ number and the $SMP^{k}$ number of complete graphs, bipartite graphs and arrangement graphs are obtained, respectively.

math.CO↗

arXiv2023

Entanglement entropy as an order parameter for strongly coupled nodal line semimetals

Topological semimetals are a class of many-body systems exhibiting novel macroscopic quantum phenomena at the interplay between high energy and condensed matter physics. They display a topological quantum phase transition (TQPT) which evades the standard Landau paradigm. In the case of Weyl semimetals, the anomalous Hall effect is a good non-local order parameter for the TQPT, as it is proportional to the separation between the Weyl nodes in momentum space. On the contrary, for nodal line semimetals (NLSM), the quest for an order parameter is still open. By taking advantage of a recently proposed holographic model for strongly-coupled NLSM, we explicitly show that entanglement entropy (EE) provides an optimal probe for nodal topology. We propose a generalized $c$-function, constructed from the EE, as an order parameter for the TQPT. Moreover, we find that the derivative of the renormalized EE with respect to the external coupling driving the TQPT diverges at the critical point, signaling the rise of non-local quantum correlations. Finally, we show that these quantum information quantities might be able to characterize not only the critical point but the whole quantum critical region at finite temperature.

hep-th↗

arXiv2023

Uncovering and Quantifying Social Biases in Code Generation

With the popularity of automatic code generation tools, such as Copilot, the study of the potential hazards of these tools is gaining importance. In this work, we explore the social bias problem in pre-trained code generation models. We propose a new paradigm to construct code prompts and successfully uncover social biases in code generation models. To quantify the severity of social biases in generated code, we develop a dataset along with three metrics to evaluate the overall social bias and fine-grained unfairness across different demographics. Experimental results on three pre-trained code generation models (Codex, InCoder, and CodeGen) with varying sizes, reveal severe social biases. Moreover, we conduct analysis to provide useful insights for further choice of code generation models with low social bias. (This work contains examples that potentially implicate stereotypes, associations, and other harms that could be offensive to individuals in certain social groups.)

cs.CL↗

arXiv2023

The statistical nature of h-index of a network node

Evaluating the importance of a network node is a crucial task in network science and graph data mining. H-index is a popular centrality measure for this task, however, there is still a lack of its interpretation from a rigorous statistical aspect. Here we show the statistical nature of h-index from the perspective of order statistics, and we obtain a new family of centrality indices by generalizing the h-index along this direction. The theoretical and empirical evidences show that such a statistical interpretation enables us to obtain a general and versatile framework for quantifying the importance of a network node. Under this framework, many new centrality indices can be derived and some of which can be more accurate and robust than h-index. We believe that this research opens up new avenues for developing more effective indices for node importance quantification from a viewpoint that still remains unexplored.

physics.soc-ph↗

arXiv2023

Spin-phonon scattering-induced low thermal conductivity in a van der Waals layered ferromagnet Cr$_2$Si$_2$Te$_6$

Layered van der Waals (vdW) magnets are prominent playgrounds for developing magnetoelectric, magneto-optic and spintronic devices. In spintronics, particularly in spincaloritronic applications, low thermal conductivity ($κ$) is highly desired. Here, by combining thermal transport measurements with density functional theory calculations, we demonstrate low $κ$ down to 1 W m$^{-1}$ K$^{-1}$ in a typical vdW ferromagnet Cr$_2$Si$_2$Te$_6$. In the paramagnetic state, development of magnetic fluctuations way above $T_\mathrm{c}=$ 33 K strongly reduces $κ$ via spin-phonon scattering, leading to low $κ\sim$ 1 W m$^{-1}$ K$^{-1}$ over a wide temperature range, in comparable to that of amorphous silica. In the magnetically ordered state, emergence of resonant magnon-phonon scattering limits $κ$ below $\sim$ 2 W m$^{-1}$ K$^{-1}$, which would be three times larger if magnetic scatterings were absent. Application of magnetic fields strongly suppresses the spin-phonon scattering, giving rise to large enhancements of $κ$. Our calculations well capture these complex behaviours of $κ$ by taking the temperature- and magnetic-field-dependent spin-phonon scattering into account. Realization of low $κ$ which is easily tunable by magnetic fields in Cr$_2$Si$_2$Te$_6$, may further promote spincaloritronic applications of vdW magnets. Our theoretical approach may also provide a generic understanding of spin-phonon scattering, which appears to play important roles in various systems.

cond-mat.mtrl-sci↗

arXiv2023

Electrical and thermal transport properties of kagome metals $A$Ti$_3$Bi$_5$ ($A$ = Rb, Cs)

We report electrical and thermal transport properties of single crystalline kagome metals $A$Ti$_3$Bi$_5$ ($A$ = Rb, Cs). Different from the structural similar kagome superconductors $A$V$_3$Sb$_5$, no charge density wave instabilities are found in $A$Ti$_3$Bi$_5$. At low temperatures below 5 K, signatures of superconductivity appear in $A$Ti$_3$Bi$_5$ as seen in magnetization measurements. However, bulk superconductivity is not evidenced by specific heat results. Similar to $A$V$_3$Sb$_5$, $A$Ti$_3$Bi$_5$ show nonlinear magnetic field dependence of the Hall effect below about 70 K, pointing to a multiband nature. Unlike $A$V$_3$Sb$_5$ in which phonons and electron-phonon coupling play important roles in thermal transport, the thermal conductivity in $A$Ti$_3$Bi$_5$ is dominated by electronic contributions. Moreover, our calculated electronic structure suggests that van Hove singularities are sitting well above the Fermi energy. Compared with $A$V$_3$Sb$_5$, the absence of charge orders in $A$Ti$_3$Bi$_5$ is closely associated with minor contributions from electron-phonon coupling and/or van Hove singularities.

cond-mat.supr-con↗

arXiv2023

Robustness of Complex Networks Considering Load and Cascading Failure under Edge-removal Attack

In the understanding of important edges in complex networks, the edges with larger degree are naturally considered more important, and they will cause greater destructiveness when attacked. However, through simulation analysis, we conclude that this understanding needs to be based on certain preconditions. In this article, the robustness of BA scale-free network and WS small-world network is studied based on the edge-removal attack strategy considering the edge load and cascading failure. Specific attack methods include: High Load Edge-removal Attacks (HLEA) and Low Load Edge-removal Attacks (LLEA). The simulation results show that the importance of edges is closely related to the load parameter δ. When 0<δ<1, attacking the edge with smaller degree will cause greater cascading failure. For this condition, the edge with smaller degree is more important. When δ=1, cascading failure is basically independent of the edge degree. When δ>1, attacking the edge with larger degree will cause greater cascading failure. Therefore, the edge with larger degree is more important for this condition.

cs.NI↗

arXiv2023

Properties of gapped systems in AdS/BCFT

We study the conductivities and entanglement structures of two different holographic gapped systems at zero density in the presence of boundaries within AdS/BCFT. The first gapped system is described by the Einstein-scalar gravity and the second one is the dual of AdS soliton geometry. We show that in both these two systems the bulk and boundary conductivities along the spatial direction of the boundary of BCFT are trivial. For the first system, when we increase the size of the subsystem the renormalized entanglement entropy is always non-negative and monotonically decreasing with discontinuous, or continuous, or smooth behavior, depending on the effective tension of the brane. While for the AdS soliton with a boundary, the renormalized entanglement entropy only exhibits a discontinuous drop when we increase the size of the subsystem.

hep-th↗

arXiv2023

Black hole interiors in holographic topological semimetals

We study the black hole interiors in holographic Weyl semimetals and holographic nodal line semimetals. We find that the black hole singularities are of Kasner form. In the topologically nontrivial phase at low temperature, both the Kasner exponents of the metric fields and the proper time from the horizon to the singularity are almost constant, likely reflecting the topological nature of the topological semimetals. We also find some specific behaviors inside the horizon in each holographic semimetal model.

hep-th↗

arXiv2023

Estimating Treatment Effects from Irregular Time Series Observations with Hidden Confounders

Causal analysis for time series data, in particular estimating individualized treatment effect (ITE), is a key task in many real-world applications, such as finance, retail, healthcare, etc. Real-world time series can include large-scale, irregular, and intermittent time series observations, raising significant challenges to existing work attempting to estimate treatment effects. Specifically, the existence of hidden confounders can lead to biased treatment estimates and complicate the causal inference process. In particular, anomaly hidden confounders which exceed the typical range can lead to high variance estimates. Moreover, in continuous time settings with irregular samples, it is challenging to directly handle the dynamics of causality. In this paper, we leverage recent advances in Lipschitz regularization and neural controlled differential equations (CDE) to develop an effective and scalable solution, namely LipCDE, to address the above challenges. LipCDE can directly model the dynamic causal relationships between historical data and outcomes with irregular samples by considering the boundary of hidden confounders given by Lipschitz-constrained neural networks. Furthermore, we conduct extensive experiments on both synthetic and real-world datasets to demonstrate the effectiveness and scalability of LipCDE.

cs.LG↗

arXiv2023

Anomalous Nernst effect in a ferrimagnetic nodal-line semiconductor Mn$_3$Si$_2$Te$_6$

In the ferrimagnetic nodal-line semiconductor Mn$_3$Si$_2$Te$_6$, colossal magnetoresistance (CMR) arises below $T_\mathrm{c}=78$ K due to the interplay of magnetism and topological nodal-line fermiology. The Berry curvature associated with the topological nodal-line is expected to produce an anomalous Nernst effect. Here, we present sizable anomalous Nernst signal in Mn$_3$Si$_2$Te$_6$ below $T_\mathrm{c}$. In the low-magnetic-field region where CMR is most apparent, the scaling ratio between the Nernst signal and magnetization is significantly enhanced compared to that in conventional magnetic materials. The enhanced Nernst effect and CMR likely share the same mechanisms, which are closely linked to the nodal-line topology.

cond-mat.str-el↗

arXiv2023

A novel efficient Multi-view traffic-related object detection framework

With the rapid development of intelligent transportation system applications, a tremendous amount of multi-view video data has emerged to enhance vehicle perception. However, performing video analytics efficiently by exploiting the spatial-temporal redundancy from video data remains challenging. Accordingly, we propose a novel traffic-related framework named CEVAS to achieve efficient object detection using multi-view video data. Briefly, a fine-grained input filtering policy is introduced to produce a reasonable region of interest from the captured images. Also, we design a sharing object manager to manage the information of objects with spatial redundancy and share their results with other vehicles. We further derive a content-aware model selection policy to select detection methods adaptively. Experimental results show that our framework significantly reduces response latency while achieving the same detection accuracy as the state-of-the-art methods.

cs.CV↗