arXiv subjects

Hao Zhang

Publications and source records attributed to Hao Zhang.

At least 757 records · Page 42Linked to original sources

arXiv2023

BRICS: Bi-level feature Representation of Image CollectionS

We present BRICS, a bi-level feature representation for image collections, which consists of a key code space on top of a feature grid space. Specifically, our representation is learned by an autoencoder to encode images into continuous key codes, which are used to retrieve features from groups of multi-resolution feature grids. Our key codes and feature grids are jointly trained continuously with well-defined gradient flows, leading to high usage rates of the feature grids and improved generative modeling compared to discrete Vector Quantization (VQ). Differently from existing continuous representations such as KL-regularized latent codes, our key codes are strictly bounded in scale and variance. Overall, feature encoding by BRICS is compact, efficient to train, and enables generative modeling over key codes using the diffusion model. Experimental results show that our method achieves comparable reconstruction results to VQ while having a smaller and more efficient decoder network (50% fewer GFlops). By applying the diffusion model over our key code space, we achieve state-of-the-art performance on image synthesis on the FFHQ and LSUN-Church (29% lower than LDM, 32% lower than StyleGAN2, 44% lower than Projected GAN on CLIP-FID) datasets.

cs.CV↗

arXiv2023

NeuralKalman: A Learnable Kalman Filter for Acoustic Echo Cancellation

The robustness of the Kalman filter to double talk and its rapid convergence make it a popular approach for addressing acoustic echo cancellation (AEC) challenges. However, the inability to model nonlinearity and the need to tune control parameters cast limitations on such adaptive filtering algorithms. In this paper, we integrate the frequency domain Kalman filter (FDKF) and deep neural networks (DNNs) into a hybrid method, called NeuralKalman, to leverage the advantages of deep learning and adaptive filtering algorithms. Specifically, we employ a DNN to estimate nonlinearly distorted far-end signals, a transition factor, and the nonlinear transition function in the state equation of the FDKF algorithm. Experimental results show that the proposed NeuralKalman improves the performance of FDKF significantly and outperforms strong baseline methods.

eess.AS↗

arXiv2023

On derivatives of zeta and $L$-functions near the 1-line

We study the conditional upper bounds and extreme values of derivatives of the Riemann zeta function and Dirichlet $L$-functions near the 1-line. Let $\ell$ be a fixed natural number. We show that, if $|σ-1|\ll1/\log_2t$, then $|ζ^{(\ell)}(σ+ it)|$ has the same maximal order (up to the leading coefficients) as $|ζ^{(\ell)}(1+ it)|$ when $t\to\infty$. The range $1-σ\ll1/\log_2t$ is wide enough, since we also show that $(1-σ) \log_2t \to \infty\; (t \to \infty)$ implies $\limsup_{t\to\infty}|ζ^{(\ell)}(σ+ it)| / (\log_2t)^{\ell+1} = \infty$. Similar results can be obtained for Dirichlet $L$-functions $L^{(\ell)}(σ,χ)$ with $χ\pmod q$.

math.NT↗

arXiv2023

Interference of cavity light by a single atom acting as a double slit

Young's double-slit interference experiment is central to quantum mechanics. While it has been demonstrated that an array of atoms can produce interference in light, it is a fundamental question to ask whether a single atom can act as a double slit when prepared in a superposition of two separate positions. Cohen-Tannoudji et al. [Proceedings of the Tenth International Conference on Laser Spectroscopy, edited by M. Ducloy, E. Giacobino, and G. Camy (World Scientific, Singapore, 1992), pp. 3-14] showed that the cross section of the light scattered by a single atom is independent of the spatial separation. In this work, however, we show that when a single atom tunneling in a double well is coupled to an optical ring cavity, the interference phenomena arise if the tunneling rate is comparable to the cavity linewidth. Being driven by an external laser in the dispersive regime, the field emitted by the atom into the cavity exhibits an interference pattern when varying the double-well spacing. Super-Poissonian bunched light can also be generated near the destructive interference. Furthermore, we show that the atomic flux of the coherent tunneling motion generates directional cavity emission, which oscillates for many cycles before the decoherence of the atomic motion and the decay of the cavity photons. Our work opens ways to manipulate photons with controllable external states of atoms for quantum information applications and use cavity light as nondestructive measurements for many-body states of atomic systems.

quant-ph↗

arXiv2023

Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena

Evaluating large language model (LLM) based chat assistants is challenging due to their broad capabilities and the inadequacy of existing benchmarks in measuring human preferences. To address this, we explore using strong LLMs as judges to evaluate these models on more open-ended questions. We examine the usage and limitations of LLM-as-a-judge, including position, verbosity, and self-enhancement biases, as well as limited reasoning ability, and propose solutions to mitigate some of them. We then verify the agreement between LLM judges and human preferences by introducing two benchmarks: MT-bench, a multi-turn question set; and Chatbot Arena, a crowdsourced battle platform. Our results reveal that strong LLM judges like GPT-4 can match both controlled and crowdsourced human preferences well, achieving over 80% agreement, the same level of agreement between humans. Hence, LLM-as-a-judge is a scalable and explainable way to approximate human preferences, which are otherwise very expensive to obtain. Additionally, we show our benchmark and traditional benchmarks complement each other by evaluating several variants of LLaMA and Vicuna. The MT-bench questions, 3K expert votes, and 30K conversations with human preferences are publicly available at https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge.

cs.CL↗

arXiv2023

Microresonator soliton frequency combs via cascaded Brillouin scattering

We demonstrate Kerr soliton frequency comb generation that is seeded by a cascaded Brillouin scattering process. In this process, a pump laser is used to generate multiple orders of Brillouin sidebands in a microresonator, which in turn generate the soliton. In such a process, even orders of Brillouin scattering sidebands are co-propagating with respect to the pump laser while odd orders of Brillouin scattering are backwards propagating. In this work we present the generation of forward propagating Kerr solitons via a forward propagating second order Brillouin scattering process in a fused silica rod resonator. Importantly, we show that the Brillouin scattering process can bridge the gap between different microresonator mode families, such that the repetition rate of the Kerr soliton is independent from the Brillouin gain frequency shift (about 10 GHz in fused silica). In our work we demonstrate this by generating soliton pulse trains with a repetition rate of 107 GHz. Our work opens up a new way for using cascaded Brillouin lasing as a seed for microresonator frequency comb generation. This can be of particular interest for the realization of soliton frequency combs with low noise properties from Brillouin lasing while still having arbitrary repetition rates that are determined by the resonator size. Applications range from optical communication to LIDAR systems and photonic signal generation.

physics.optics↗

arXiv2023

LARP: Language-Agent Role Play for Open-World Games

Language agents have shown impressive problem-solving skills within defined settings and brief timelines. Yet, with the ever-evolving complexities of open-world simulations, there's a pressing need for agents that can flexibly adapt to complex environments and consistently maintain a long-term memory to ensure coherent actions. To bridge the gap between language agents and open-world games, we introduce Language Agent for Role-Playing (LARP), which includes a cognitive architecture that encompasses memory processing and a decision-making assistant, an environment interaction module with a feedback-driven learnable action space, and a postprocessing method that promotes the alignment of various personalities. The LARP framework refines interactions between users and agents, predefined with unique backgrounds and personalities, ultimately enhancing the gaming experience in open-world contexts. Furthermore, it highlights the diverse uses of language models in a range of areas such as entertainment, education, and various simulation scenarios. The project page is released at https://miao-ai-lab.github.io/LARP/.

cs.AI↗

arXiv2023

Compositional Zero-Shot Learning for Attribute-Based Object Reference in Human-Robot Interaction

Language-enabled robots have been widely studied over the past years to enable natural human-robot interaction and teaming in various real-world applications. Language-enabled robots must be able to comprehend referring expressions to identify a particular object from visual perception using a set of referring attributes extracted from natural language. However, visual observations of an object may not be available when it is referred to, and the number of objects and attributes may also be unbounded in open worlds. To address the challenges, we implement an attribute-based compositional zero-shot learning method that uses a list of attributes to perform referring expression comprehension in open worlds. We evaluate the approach on two datasets including the MIT-States and the Clothing 16K. The preliminary experimental results show that our implemented approach allows a robot to correctly identify the objects referred to by human commands.

cs.RO↗

arXiv2023

De novo Drug Design using Reinforcement Learning with Multiple GPT Agents

De novo drug design is a pivotal issue in pharmacology and a new area of focus in AI for science research. A central challenge in this field is to generate molecules with specific properties while also producing a wide range of diverse candidates. Although advanced technologies such as transformer models and reinforcement learning have been applied in drug design, their potential has not been fully realized. Therefore, we propose MolRL-MGPT, a reinforcement learning algorithm with multiple GPT agents for drug molecular generation. To promote molecular diversity, we encourage the agents to collaborate in searching for desirable molecules in diverse directions. Our algorithm has shown promising results on the GuacaMol benchmark and exhibits efficacy in designing inhibitors against SARS-CoV-2 protein targets. The codes are available at: https://github.com/HXYfighter/MolRL-MGPT.

q-bio.BM↗

arXiv2023

In situ tuning of dynamical Coulomb blockade on Andreev bound states in hybrid nanowire devices

Electron interactions in quantum devices can exhibit intriguing phenomena. One example is assembling an electronic device in series with an on-chip resistor. The quantum laws of electricity of the device is modified at low energies and temperatures by dissipative interactions induced by the resistor, a phenomenon known as dynamical Coulomb blockade (DCB). The DCB strength is usually non-adjustable in a fixed environment defined by the resistor. Here, we design an on-chip circuit for InAs-Al hybrid nanowires where the DCB strength can be gate-tuned in situ. InAs-Al nanowires could host Andreev or Majorana zero-energy states. This technique enables tracking the evolution of the same state while tuning the DCB strength from weak to strong. We observe the transition from a zero-bias conductance peak to split peaks for Andreev zero-energy states. Our technique opens the door to in situ tuning interaction strength on zero-energy states.

cond-mat.mes-hall↗

arXiv2023

Combined Invariant Subspace \& Frequency-Domain Subspace Method for Identification of Discrete-Time MIMO Linear Systems

Recently, a novel system identification method based on invariant subspace theory is introduced, aiming to address the identification problem of continuous-time (CT) linear time-invariant (LTI) systems by combining time-domain and frequency-domain methods. Subsequently, the combined Invariant-Subspace and Subspace Identification Method (cISSIM) is introduced, enabling direct estimation of CT LTI systems in state-space forms. It produces consistent estimation that is robust in an error-in-variable and slow-sampling conditions, while no pre-filtering operation of the input-output signals is needed. This paper presents the discrete-cISSIM, which extends cISSIM to discrete-time (DT) systems and offers the following improvements: 1) the capability to utilize arbitrary discrete periodic excitations while cISSIM uses multi-sine signals; 2) a faster estimation with reduced computational complexity is proposed; 3) the covariance estimation problem can be addressed concurrently with the system parameter estimation. An implementation of discrete-cISSIM by MATLAB has also been provided.

eess.SY↗

arXiv2023

Anatomy of topological Anderson transitions

We study mesoscopic signatures of the topological Anderson transitions in topological disordered chains. To this end we introduce an integer-valued sample-specific definition of the topological index in finite size systems. Its phase diagram exhibits a fascinating structure of intermittent topological phases, dubbed topological islands. Their existence is rooted in the real zeros of the underlying random polynomial. Their statistics exhibits finite-size scaling, pointing to the location of the bulk topological Anderson transition. While the average theories in AIII and BDI symmetry classes are rather similar, the corresponding patterns of topological islands and their statistics are qualitatively different. We also discuss observable signatures of sharp topological transitions in mesoscopic systems, such as persistent currents and entanglement spectra.

cond-mat.dis-nn↗

arXiv2023

Slice3D: Multi-Slice, Occlusion-Revealing, Single View 3D Reconstruction

We introduce multi-slice reasoning, a new notion for single-view 3D reconstruction which challenges the current and prevailing belief that multi-view synthesis is the most natural conduit between single-view and 3D. Our key observation is that object slicing is more advantageous than altering views to reveal occluded structures. Specifically, slicing is more occlusion-revealing since it can peel through any occluders without obstruction. In the limit, i.e., with infinitely many slices, it is guaranteed to unveil all hidden object parts. We realize our idea by developing Slice3D, a novel method for single-view 3D reconstruction which first predicts multi-slice images from a single RGB image and then integrates the slices into a 3D model using a coordinate-based transformer network for signed distance prediction. The slice images can be regressed or generated, both through a U-Net based network. For the former, we inject a learnable slice indicator code to designate each decoded image into a spatial slice location, while the slice generator is a denoising diffusion model operating on the entirety of slice images stacked on the input channels. We conduct extensive evaluation against state-of-the-art alternatives to demonstrate superiority of our method, especially in recovering complex and severely occluded shape structures, amid ambiguities. All Slice3D results were produced by networks trained on a single Nvidia A40 GPU, with an inference time less than 20 seconds.

cs.CV↗

arXiv2023

Neural Eigenfunctions Are Structured Representation Learners

This paper introduces a structured, adaptive-length deep representation called Neural Eigenmap. Unlike prior spectral methods such as Laplacian Eigenmap that operate in a nonparametric manner, Neural Eigenmap leverages NeuralEF to parametrically model eigenfunctions using a neural network. We show that, when the eigenfunction is derived from positive relations in a data augmentation setup, applying NeuralEF results in an objective function that resembles those of popular self-supervised learning methods, with an additional symmetry-breaking property that leads to \emph{structured} representations where features are ordered by importance. We demonstrate using such representations as adaptive-length codes in image retrieval systems. By truncation according to feature importance, our method requires up to $16\times$ shorter representation length than leading self-supervised learning ones to achieve similar retrieval performance. We further apply our method to graph data and report strong results on a node representation learning benchmark with more than one million nodes.

cs.LG↗

arXiv2023

Knowledge Unlearning for LLMs: Tasks, Methods, and Challenges

In recent years, large language models (LLMs) have spurred a new research paradigm in natural language processing. Despite their excellent capability in knowledge-based question answering and reasoning, their potential to retain faulty or even harmful knowledge poses risks of malicious application. The challenge of mitigating this issue and transforming these models into purer assistants is crucial for their widespread applicability. Unfortunately, Retraining LLMs repeatedly to eliminate undesirable knowledge is impractical due to their immense parameters. Knowledge unlearning, derived from analogous studies on machine unlearning, presents a promising avenue to address this concern and is notably advantageous in the context of LLMs. It allows for the removal of harmful knowledge in an efficient manner, without affecting unrelated knowledge in the model. To this end, we provide a survey of knowledge unlearning in the era of LLMs. Firstly, we formally define the knowledge unlearning problem and distinguish it from related works. Subsequently, we categorize existing knowledge unlearning methods into three classes: those based on parameter optimization, parameter merging, and in-context learning, and introduce details of these unlearning methods. We further present evaluation datasets used in existing methods, and finally conclude this survey by presenting the ongoing challenges and future directions.

cs.CL↗

arXiv2023

LLaVA-Grounding: Grounded Visual Chat with Large Multimodal Models

With the recent significant advancements in large multi-modal models (LMMs), the importance of their grounding capability in visual chat is increasingly recognized. Despite recent efforts to enable LMMs to support grounding, their capabilities for grounding and chat are usually separate, and their chat performance drops dramatically when asked to ground. The problem is the lack of a dataset for grounded visual chat (GVC). Existing grounding datasets only contain short captions. To address this issue, we have created GVC data that allows for the combination of grounding and chat capabilities. To better evaluate the GVC capabilities, we have introduced a benchmark called Grounding-Bench. Additionally, we have proposed a model design that can support GVC and various types of visual prompts by connecting segmentation models with language models. Experimental results demonstrate that our model outperforms other LMMs on Grounding-Bench. Furthermore, our model achieves competitive performance on classic grounding benchmarks like RefCOCO/+/g and Flickr30K Entities. Our code will be released at https://github.com/UX-Decoder/LLaVA-Grounding .

cs.CV↗

arXiv2023

FaceDNeRF: Semantics-Driven Face Reconstruction, Prompt Editing and Relighting with Diffusion Models

The ability to create high-quality 3D faces from a single image has become increasingly important with wide applications in video conferencing, AR/VR, and advanced video editing in movie industries. In this paper, we propose Face Diffusion NeRF (FaceDNeRF), a new generative method to reconstruct high-quality Face NeRFs from single images, complete with semantic editing and relighting capabilities. FaceDNeRF utilizes high-resolution 3D GAN inversion and expertly trained 2D latent-diffusion model, allowing users to manipulate and construct Face NeRFs in zero-shot learning without the need for explicit 3D data. With carefully designed illumination and identity preserving loss, as well as multi-modal pre-training, FaceDNeRF offers users unparalleled control over the editing process enabling them to create and edit face NeRFs using just single-view images, text prompts, and explicit target lighting. The advanced features of FaceDNeRF have been designed to produce more impressive results than existing 2D editing approaches that rely on 2D segmentation maps for editable attributes. Experiments show that our FaceDNeRF achieves exceptionally realistic results and unprecedented flexibility in editing compared with state-of-the-art 3D face reconstruction and editing methods. Our code will be available at https://github.com/BillyXYB/FaceDNeRF.

cs.CV↗

arXiv2023

Autonomous and Adaptive Role Selection for Multi-robot Collaborative Area Search Based on Deep Reinforcement Learning

In the tasks of multi-robot collaborative area search, we propose the unified approach for simultaneous mapping for sensing more targets (exploration) while searching and locating the targets (coverage). Specifically, we implement a hierarchical multi-agent reinforcement learning algorithm to decouple task planning from task execution. The role concept is integrated into the upper-level task planning for role selection, which enables robots to learn the role based on the state status from the upper-view. Besides, an intelligent role switching mechanism enables the role selection module to function between two timesteps, promoting both exploration and coverage interchangeably. Then the primitive policy learns how to plan based on their assigned roles and local observation for sub-task execution. The well-designed experiments show the scalability and generalization of our method compared with state-of-the-art approaches in the scenes with varying complexity and number of robots.

cs.RO↗