arXiv subjects

Qi Guo

Publications and source records attributed to Qi Guo.

At least 199 records · Page 11Linked to original sources

arXiv2023

Neutral Hydrogen content of dwarf galaxies in different environments

Environments play an important role in galaxy formation and evolution, particularly in regulating the content of neutral gas. However, current HI surveys have limitations in their depth, which prevents them from adequately studying low HI content galaxies in high-density regions. In this study, we address this issue by employing the Five-hundred-meter Aperture Spherical radio Telescope (FAST) with extensive integration times to complement the relatively shallow Arecibo Legacy Fast Arecibo L-band Feed Array (ALFALFA) HI survey. This approach allows us to explore the gas content of dwarf galaxies across various environments. We observe a positive relationship between HI mass and stellar mass in dwarf galaxies, with a well-defined upper boundary for HI mass that holds true in both observations and simulations. Furthermore, we find a decrease in the HI-to-stellar mass ratio ($\rm M_{\rm HI}/M_*$) as the density of the environment increases, irrespective of whether it is determined by the proximity to the nearest group or the projected number density. Comparing our observations to simulations, we note a steeper slope in the relationship, indicating a gradual gas-stripping process in the observational data. Additionally, we find that the scaling relation between the $\rm M_{\rm HI}/M_*$ and optical properties can be improved by incorporating galaxy environments.

astro-ph.GA↗

arXiv2023

Self-driven Grounding: Large Language Model Agents with Automatical Language-aligned Skill Learning

Large language models (LLMs) show their powerful automatic reasoning and planning capability with a wealth of semantic knowledge about the human world. However, the grounding problem still hinders the applications of LLMs in the real-world environment. Existing studies try to fine-tune the LLM or utilize pre-defined behavior APIs to bridge the LLMs and the environment, which not only costs huge human efforts to customize for every single task but also weakens the generality strengths of LLMs. To autonomously ground the LLM onto the environment, we proposed the Self-Driven Grounding (SDG) framework to automatically and progressively ground the LLM with self-driven skill learning. SDG first employs the LLM to propose the hypothesis of sub-goals to achieve tasks and then verify the feasibility of the hypothesis via interacting with the underlying environment. Once verified, SDG can then learn generalized skills with the guidance of these successfully grounded subgoals. These skills can be further utilized to accomplish more complex tasks which fail to pass the verification phase. Verified in the famous instruction following task set-BabyAI, SDG achieves comparable performance in the most challenging tasks compared with imitation learning methods that cost millions of demonstrations, proving the effectiveness of learned skills and showing the feasibility and efficiency of our framework.

cs.CL↗

arXiv2023

Existence and Multiplicity of Normalized Solutions for Dirac Equations with non-autonomous nonlinearities

In this paper, we study the following nonlinear Dirac equations \begin{align*} \begin{cases} -i\sum\limits_{k=1}^3α_k\partial_k u+mβu=f(x,|u|)u+ωu, \displaystyle \int_{\mathbb{R}^3} |u|^2dx=a^2, \end{cases} \end{align*} where $u: \mathbb{R}^{3}\rightarrow \mathbb{C}^{4}$, $m>0$ is the mass of the Dirac particle, $ω\in \mathbb{R}$ arises as a Lagrange multiplier, $\partial_k=\frac{\partial}{\partial x_k}$, $α_1,α_2,α_3$ are $4\times 4$ Pauli-Dirac matrices, $a>0$ is a prescribed constant, and $f(x,\cdot)$ has several physical interpretations that will be discussed in the Introduction. Under general assumptions on the nonlinearity $f$, we prove the existence of $L^2$-normalized solutions for the above nonlinear Dirac equations by using perturbation methods in combination with Lyapunov-Schmidt reduction. We also show the multiplicity of these normalized solutions thanks to the multiplicity theorem of Ljusternik-Schnirelmann. Moreover, we obtain bifurcation results of this problem.

math.AP↗

arXiv2023

Conditional Colour-Magnitude Distribution of Central Galaxies in Galaxy Formation Models

We investigate the conditional colour-magnitude distribution (CCMD), namely the colour-magnitude distribution at fixed halo mass, of the central galaxies in semi-analytic galaxy formation model (SAM) and hydrodynamic simulations. We analyse the CCMD of central galaxies in each halo mass bin with the Gaussian mixture model and find that it can be decomposed into red and blue components nearly orthogonal to each other, a red component narrow in colour and extended in magnitude and a blue component narrow in magnitude and extended in colour. We focus on the SAM galaxies to explore the origin of the CCMD components by studying the relation between central galaxy colour and halo or galaxy properties. Central galaxy colour is correlated with halo assembly properties for low mass haloes and independent of them for high mass haloes. Galaxy properties such as central supermassive black hole mass, cold gas mass, and gas specific angular momentum can all impact central galaxy colour. These results are corroborated by an alternative machine learning analysis in which we attempt to predict central galaxy colour with halo and galaxy properties. We find that the prediction for colours of central galaxies can be significantly improved using both halo and galaxy properties as input compared to using halo properties alone. With the halo and galaxy properties considered here, we find that subtle discrepancies remain between predicted and original colour distribution for low mass haloes and that no significant determining properties are identified in massive haloes, suggesting modulations by additional stochastic processes in galaxy formation.

astro-ph.GA↗

arXiv2023

Pushing the Limits of Machine Design: Automated CPU Design with AI

Design activity -- constructing an artifact description satisfying given goals and constraints -- distinguishes humanity from other animals and traditional machines, and endowing machines with design abilities at the human level or beyond has been a long-term pursuit. Though machines have already demonstrated their abilities in designing new materials, proteins, and computer programs with advanced artificial intelligence (AI) techniques, the search space for designing such objects is relatively small, and thus, "Can machines design like humans?" remains an open question. To explore the boundary of machine design, here we present a new AI approach to automatically design a central processing unit (CPU), the brain of a computer, and one of the world's most intricate devices humanity have ever designed. This approach generates the circuit logic, which is represented by a graph structure called Binary Speculation Diagram (BSD), of the CPU design from only external input-output observations instead of formal program code. During the generation of BSD, Monte Carlo-based expansion and the distance of Boolean functions are used to guarantee accuracy and efficiency, respectively. By efficiently exploring a search space of unprecedented size 10^{10^{540}}, which is the largest one of all machine-designed objects to our best knowledge, and thus pushing the limits of machine design, our approach generates an industrial-scale RISC-V CPU within only 5 hours. The taped-out CPU successfully runs the Linux operating system and performs comparably against the human-designed Intel 80486SX CPU. In addition to learning the world's first CPU only from input-output observations, which may reform the semiconductor industry by significantly reducing the design cycle, our approach even autonomously discovers human knowledge of the von Neumann architecture.

cs.AI↗

arXiv2023

Online Prototype Alignment for Few-shot Policy Transfer

Domain adaptation in reinforcement learning (RL) mainly deals with the changes of observation when transferring the policy to a new environment. Many traditional approaches of domain adaptation in RL manage to learn a mapping function between the source and target domain in explicit or implicit ways. However, they typically require access to abundant data from the target domain. Besides, they often rely on visual clues to learn the mapping function and may fail when the source domain looks quite different from the target domain. To address these problems, we propose a novel framework Online Prototype Alignment (OPA) to learn the mapping function based on the functional similarity of elements and is able to achieve the few-shot policy transfer within only several episodes. The key insight of OPA is to introduce an exploration mechanism that can interact with the unseen elements of the target domain in an efficient and purposeful manner, and then connect them with the seen elements in the source domain according to their functionalities (instead of visual clues). Experimental results show that when the target domain looks visually different from the source domain, OPA can achieve better transfer performance even with much fewer samples from the target domain, outperforming prior methods.

cs.LG↗

arXiv2023

Path-Specific Counterfactual Fairness for Recommender Systems

Recommender systems (RSs) have become an indispensable part of online platforms. With the growing concerns of algorithmic fairness, RSs are not only expected to deliver high-quality personalized content, but are also demanded not to discriminate against users based on their demographic information. However, existing RSs could capture undesirable correlations between sensitive features and observed user behaviors, leading to biased recommendations. Most fair RSs tackle this problem by completely blocking the influences of sensitive features on recommendations. But since sensitive features may also affect user interests in a fair manner (e.g., race on culture-based preferences), indiscriminately eliminating all the influences of sensitive features inevitably degenerate the recommendations quality and necessary diversities. To address this challenge, we propose a path-specific fair RS (PSF-RS) for recommendations. Specifically, we summarize all fair and unfair correlations between sensitive features and observed ratings into two latent proxy mediators, where the concept of path-specific bias (PS-Bias) is defined based on path-specific counterfactual inference. Inspired by Pearl's minimal change principle, we address the PS-Bias by minimally transforming the biased factual world into a hypothetically fair world, where a fair RS model can be learned accordingly by solving a constrained optimization problem. For the technical part, we propose a feasible implementation of PSF-RS, i.e., PSF-VAE, with weakly-supervised variational inference, which robustly infers the latent mediators such that unfairness can be mitigated while necessary recommendation diversities can be maximally preserved simultaneously. Experiments conducted on semi-simulated and real-world datasets demonstrate the effectiveness of PSF-RS.

cs.IR↗

arXiv2023

Dual Class-Aware Contrastive Federated Semi-Supervised Learning

Federated semi-supervised learning (FSSL), facilitates labeled clients and unlabeled clients jointly training a global model without sharing private data. Existing FSSL methods predominantly employ pseudo-labeling and consistency regularization to exploit the knowledge of unlabeled data, achieving notable success in raw data utilization. However, these training processes are hindered by large deviations between uploaded local models of labeled and unlabeled clients, as well as confirmation bias introduced by noisy pseudo-labels, both of which negatively affect the global model's performance. In this paper, we present a novel FSSL method called Dual Class-aware Contrastive Federated Semi-Supervised Learning (DCCFSSL). This method accounts for both the local class-aware distribution of each client's data and the global class-aware distribution of all clients' data within the feature space. By implementing a dual class-aware contrastive module, DCCFSSL establishes a unified training objective for different clients to tackle large deviations and incorporates contrastive information in the feature space to mitigate confirmation bias. Moreover, DCCFSSL introduces an authentication-reweighted aggregation technique to improve the server's aggregation robustness. Our comprehensive experiments show that DCCFSSL outperforms current state-of-the-art methods on three benchmark datasets and surpasses the FedAvg with relabeled unlabeled clients on CIFAR-10, CIFAR-100, and STL-10 datasets. To our knowledge, we are the first to present an FSSL method that utilizes only 10\% labeled clients, while still achieving superior performance compared to standard federated supervised learning, which uses all clients with labeled data.

cs.LG↗

arXiv2023

Gradient estimates for positive weak solution to $Δ_pu+au^σ=0$ on Riemannian manifolds

In this paper, we study gradient estimates for positive weak solutions to the following $p$-Laplacian equation $$Δ_pu+au^σ=0$$ on a Riemannian manifold, where $p>1$ and $a,σ$ are two nonzero real constants. By virtue of the Morser iteration technique, we derive some gradient estimates, which show that when the Ricci curvature is nonnegative, the above equation does not admit positive weak solutions under some scopes of $p$.

math.AP↗

arXiv2023

Global dynamic scaling relations of HI-rich ultra-diffuse galaxies

The baryonic Tully-Fisher relation (BTFR), which connects the baryonic mass of galaxies with their circular velocities, has been validated across a wide range of galaxies, from dwarf galaxies to massive galaxies. Recent studies have found that several ultra-diffuse galaxies (UDGs) deviate significantly from the BTFR, indicating a galaxy population with abnormal dynamical properties. However, such studies were still confined within a small sample size. In this study, we used the 100% complete Arecibo Legacy Fast Arecibo L-band Feed Array (ALFALFA) to investigate the BTFR of 88 HI-rich UDGs (HUDGs), which is the largest UDG sample with dynamical information. We found that the HUDGs form a continuous distribution in the BTFR diagram, with high-velocity galaxies consistent with normal dwarf galaxies at 1 $σ$ level, and low-velocity galaxies deviating from the BTFR, in line with that reported in the literature. We point out that the observed deviation may be subject to various selection effects or systemic biases. Nevertheless, we found that the significance of the deviation of HUDGs from the BTFR and TFR are different, i.e., they either deviate from the BTFR or from the TFR. Our result indicates that a high-gas fraction may play an important role in explaining the deviation of HUDGs from BTFR.

astro-ph.GA↗

arXiv2023

The galaxy size to halo spin relation of disk galaxies in cosmological hydrodynamical simulations

In the standard disk galaxy formation model, the sizes of galactic disks are tightly related to the spin parameters $λ$ of their dark matter haloes. The model has been wildly adopted by various semi-analytic galaxy formation models which have been extremely successful to interpret a large body of observational data. However, the size-$λ$ correlation was rarely seen in most modern hydrodynamical simulations of galaxy formation. In this short paper, we make use of 4 sets of large hydrodynamical simulations to explore the size-spin parameter relation with a large sample of simulated disk galaxies, and compare it with a popular disk galaxy formation model of Mo et al. (1998). Intriguingly, galactic sizes correlate strongly with spin parameters of their dark matter haloes in the simulations developed by the IllustrisTNG collaborations, albeit the relation does not always agree with prediction of MMW98 model over all stellar mass range we examined. There is also a size-spin correlation for the Milky way analogies in the EAGLE simulations, while it is relatively weaker than that of the IllustrisTNG counterparts. For the dwarfs in the simulations from the EAGLE collaboration, there is NULL correlation. We conclude that either the detailed subgrid physics or hydrodynamics solvers account for the size-spin relation, which will be explored in our future work.

astro-ph.GA↗

arXiv2023

Conceptual Reinforcement Learning for Language-Conditioned Tasks

Despite the broad application of deep reinforcement learning (RL), transferring and adapting the policy to unseen but similar environments is still a significant challenge. Recently, the language-conditioned policy is proposed to facilitate policy transfer through learning the joint representation of observation and text that catches the compact and invariant information across environments. Existing studies of language-conditioned RL methods often learn the joint representation as a simple latent layer for the given instances (episode-specific observation and text), which inevitably includes noisy or irrelevant information and cause spurious correlations that are dependent on instances, thus hurting generalization performance and training efficiency. To address this issue, we propose a conceptual reinforcement learning (CRL) framework to learn the concept-like joint representation for language-conditioned policy. The key insight is that concepts are compact and invariant representations in human cognition through extracting similarities from numerous instances in real-world. In CRL, we propose a multi-level attention encoder and two mutual information constraints for learning compact and invariant concepts. Verified in two challenging environments, RTFM and Messenger, CRL significantly improves the training efficiency (up to 70%) and generalization ability (up to 30%) to the new environment dynamics.

cs.LG↗

arXiv2023

Dark against luminous matter around isolated central galaxies: a comparative study between modern surveys and Illustris-TNG

Based on independent shear measurements using the DECaLS/DR8 imaging data, we measure the weak lensing signals around isolated central galaxies (ICGs) from SDSS/DR7 at $z\sim0.1$. The projected stellar mass density profiles of surrounding satellite galaxies are further deduced, using photometric sources from the Hyper Suprime-Cam (HSC) survey (pDR3). The signals of ICGs $+$ their extended stellar halos are taken from Wang et al.(2021). All measurements are compared with predictions by the Illustris-TNG300-1 simulation. We find, overall, a good agreement between observation and TNG300. In particular, a correction to the stellar mass of massive observed ICGs is applied based on the calibration of He et al.(2013), which brings a much better agreement with TNG300 predicted lensing signals at $\log_{10}M_\ast/M_\odot>11.1$. In real observation, red ICGs are hosted by more massive dark matter halos, have more satellites and more extended stellar halos than blue ICGs at fixed stellar mass. However, in TNG300 there are more satellites around blue ICGs at fixed stellar mass, and the outer stellar halos of red and blue ICGs are similar. The stellar halos of TNG galaxies are more extended compared with real observed galaxies, especially for blue ICGs with $\log_{10}M_\ast/M_\odot>10.8$. We find the same trend for TNG100 galaxies and for true halo central galaxies. The tensions between TNG and real galaxies might indicate that satellite disruptions are stronger in TNG. In both TNG300 and observation, satellites approximately trace the underlying dark matter distribution beyond $0.1R_{200}$, but the fraction of total stellar mass in TNG300 does not show the same radial distribution as real galaxies.

astro-ph.GA↗

arXiv2023

Ultra-low Precision Multiplication-free Training for Deep Neural Networks

The training for deep neural networks (DNNs) demands immense energy consumption, which restricts the development of deep learning as well as increases carbon emissions. Thus, the study of energy-efficient training for DNNs is essential. In training, the linear layers consume the most energy because of the intense use of energy-consuming full-precision (FP32) multiplication in multiply-accumulate (MAC). The energy-efficient works try to decrease the precision of multiplication or replace the multiplication with energy-efficient operations such as addition or bitwise shift, to reduce the energy consumption of FP32 multiplications. However, the existing energy-efficient works cannot replace all of the FP32 multiplications during both forward and backward propagation with low-precision energy-efficient operations. In this work, we propose an Adaptive Layer-wise Scaling PoT Quantization (ALS-POTQ) method and a Multiplication-Free MAC (MF-MAC) to replace all of the FP32 multiplications with the INT4 additions and 1-bit XOR operations. In addition, we propose Weight Bias Correction and Parameterized Ratio Clipping techniques for stable training and improving accuracy. In our training scheme, all of the above methods do not introduce extra multiplications, so we reduce up to 95.8% of the energy consumption in linear layers during training. Experimentally, we achieve an accuracy degradation of less than 1% for CNN models on ImageNet and Transformer model on the WMT En-De task. In summary, we significantly outperform the existing methods for both energy efficiency and accuracy.

cs.LG↗

arXiv2023

Online Symbolic Regression with Informative Query

Symbolic regression, the task of extracting mathematical expressions from the observed data $\{ \vx_i, y_i \}$, plays a crucial role in scientific discovery. Despite the promising performance of existing methods, most of them conduct symbolic regression in an \textit{offline} setting. That is, they treat the observed data points as given ones that are simply sampled from uniform distributions without exploring the expressive potential of data. However, for real-world scientific problems, the data used for symbolic regression are usually actively obtained by doing experiments, which is an \textit{online} setting. Thus, how to obtain informative data that can facilitate the symbolic regression process is an important problem that remains challenging. In this paper, we propose QUOSR, a \textbf{qu}ery-based framework for \textbf{o}nline \textbf{s}ymbolic \textbf{r}egression that can automatically obtain informative data in an iterative manner. Specifically, at each step, QUOSR receives historical data points, generates new $\vx$, and then queries the symbolic expression to get the corresponding $y$, where the $(\vx, y)$ serves as new data points. This process repeats until the maximum number of query steps is reached. To make the generated data points informative, we implement the framework with a neural network and train it by maximizing the mutual information between generated data points and the target expression. Through comprehensive experiments, we show that QUOSR can facilitate modern symbolic regression methods by generating informative data.

cs.LG↗

arXiv2023

Quantized Distributed Training of Large Models with Convergence Guarantees

Communication-reduction techniques are a popular way to improve scalability in data-parallel training of deep neural networks (DNNs). The recent emergence of large language models such as GPT has created the need for new approaches to exploit data-parallelism. Among these, fully-sharded data parallel (FSDP) training is highly popular, yet it still encounters scalability bottlenecks. One reason is that applying compression techniques to FSDP is challenging: as the vast majority of the communication involves the model's weights, direct compression alters convergence and leads to accuracy loss. We present QSDP, a variant of FSDP which supports both gradient and weight quantization with theoretical guarantees, is simple to implement and has essentially no overheads. To derive QSDP we prove that a natural modification of SGD achieves convergence even when we only maintain quantized weights, and thus the domain over which we train consists of quantized points and is, therefore, highly non-convex. We validate this approach by training GPT-family models with up to 1.3 billion parameters on a multi-node cluster. Experiments show that QSDP preserves model accuracy, while completely removing the communication bottlenecks of FSDP, providing end-to-end speedups of up to 2.2x.

cs.LG↗

arXiv2023

Improving Dialogue Breakdown Detection with Semi-Supervised Learning

Building user trust in dialogue agents requires smooth and consistent dialogue exchanges. However, agents can easily lose conversational context and generate irrelevant utterances. These situations are called dialogue breakdown, where agent utterances prevent users from continuing the conversation. Building systems to detect dialogue breakdown allows agents to recover appropriately or avoid breakdown entirely. In this paper we investigate the use of semi-supervised learning methods to improve dialogue breakdown detection, including continued pre-training on the Reddit dataset and a manifold-based data augmentation method. We demonstrate the effectiveness of these methods on the Dialogue Breakdown Detection Challenge (DBDC) English shared task. Our submissions to the 2020 DBDC5 shared task place first, beating baselines and other submissions by over 12\% accuracy. In ablations on DBDC4 data from 2019, our semi-supervised learning methods improve the performance of a baseline BERT model by 2\% accuracy. These methods are applicable generally to any dialogue task and provide a simple way to improve model performance.

cs.CL↗

arXiv2022

The halo concentration and mass relation traced by satellite galaxies

We study the relation between halo concentration and mass (c-M relation) using the Seventh and Eighth Data Release of the Sloan Digital Sky Survey (SDSS DR7 and DR8) galaxy catalogue. Assuming that the satellite galaxies follow the distribution of dark matter, we derive the halo concentration by fitting the satellite radial profile with a Nararro Frank and White (NFW) format. The derived c-M relation covers a wide halo mass range from $10^{11.6}$ to $10^{14.1} \rm\ M_\odot$. We confirm the anti-correlation between the halo mass and concentration as predicted in cosmological simulations. Our results are in good agreement with those derived using galaxy dynamics and gravitational lensing for halos of $10^{11.6}-10^{12.9} \rm\ M_\odot$, while they are slightly lower for halos of $10^{12.9}-10^{14.1}\rm\ M_\odot$. It is because blue satellite galaxies are less concentrated, especially in the inner regions. Instead of using all satellite galaxies, red satellites could be better tracers of the underlying dark matter distribution in galaxy groups.

astro-ph.GA↗