Searcharxiv⌕ Search

arXiv subjects

Ziyu Wang

Publications and source records attributed to Ziyu Wang.

At least 127 records · Page 7Linked to original sources
arXiv2024

ChatMusician: Understanding and Generating Music Intrinsically with LLM

While Large Language Models (LLMs) demonstrate impressive capabilities in text generation, we find that their ability has yet to be generalized to music, humanity's creative language. We introduce ChatMusician, an open-source LLM that integrates intrinsic musical abilities. It is based on continual pre-training and finetuning LLaMA2 on a text-compatible music representation, ABC notation, and the music is treated as a second language. ChatMusician can understand and generate music with a pure text tokenizer without any external multi-modal neural structures or tokenizers. Interestingly, endowing musical abilities does not harm language abilities, even achieving a slightly higher MMLU score. Our model is capable of composing well-structured, full-length music, conditioned on texts, chords, melodies, motifs, musical forms, etc, surpassing GPT-4 baseline. On our meticulously curated college-level music understanding benchmark, MusicTheoryBench, ChatMusician surpasses LLaMA2 and GPT-3.5 on zero-shot setting by a noticeable margin. Our work reveals that LLMs can be an excellent compressor for music, but there remains significant territory to be conquered. We release our 4B token music-language corpora MusicPile, the collected MusicTheoryBench, code, model and demo in GitHub.

cs.SD↗
arXiv2023

Generalizable Visual Reinforcement Learning with Segment Anything Model

Learning policies that can generalize to unseen environments is a fundamental challenge in visual reinforcement learning (RL). While most current methods focus on acquiring robust visual representations through auxiliary supervision, pre-training, or data augmentation, the potential of modern vision foundation models remains underleveraged. In this work, we introduce Segment Anything Model for Generalizable visual RL (SAM-G), a novel framework that leverages the promptable segmentation ability of Segment Anything Model (SAM) to enhance the generalization capabilities of visual RL agents. We utilize image features from DINOv2 and SAM to find correspondence as point prompts to SAM, and then SAM produces high-quality masked images for agents directly. Evaluated across 8 DMControl tasks and 3 Adroit tasks, SAM-G significantly improves the visual generalization ability without altering the RL agents' architecture but merely their observations. Notably, SAM-G achieves 44% and 29% relative improvements on the challenging video hard setting on DMControl and Adroit respectively, compared to state-of-the-art methods. Video and code: https://yanjieze.com/SAM-G/

cs.LG↗
arXiv2023

MVHuman: Tailoring 2D Diffusion with Multi-view Sampling For Realistic 3D Human Generation

Recent months have witnessed rapid progress in 3D generation based on diffusion models. Most advances require fine-tuning existing 2D Stable Diffsuions into multi-view settings or tedious distilling operations and hence fall short of 3D human generation due to the lack of diverse 3D human datasets. We present an alternative scheme named MVHuman to generate human radiance fields from text guidance, with consistent multi-view images directly sampled from pre-trained Stable Diffsuions without any fine-tuning or distilling. Our core is a multi-view sampling strategy to tailor the denoising processes of the pre-trained network for generating consistent multi-view images. It encompasses view-consistent conditioning, replacing the original noises with ``consistency-guided noises'', optimizing latent codes, as well as utilizing cross-view attention layers. With the multi-view images through the sampling process, we adopt geometry refinement and 3D radiance field generation followed by a subsequent neural blending scheme for free-view rendering. Extensive experiments demonstrate the efficacy of our method, as well as its superiority to state-of-the-art 3D human generation methods.

cs.CV↗
arXiv2023

RigLSTM: Recurrent Independent Grid LSTM for Generalizable Sequence Learning

Sequential processes in real-world often carry a combination of simple subsystems that interact with each other in certain forms. Learning such a modular structure can often improve the robustness against environmental changes. In this paper, we propose recurrent independent Grid LSTM (RigLSTM), composed of a group of independent LSTM cells that cooperate with each other, for exploiting the underlying modular structure of the target task. Our model adopts cell selection, input feature selection, hidden state selection, and soft state updating to achieve a better generalization ability on the basis of the recent Grid LSTM for the tasks where some factors differ between training and evaluation. Specifically, at each time step, only a fraction of cells are activated, and the activated cells select relevant inputs and cells to communicate with. At the end of one time step, the hidden states of the activated cells are updated by considering the relevance between the inputs and the hidden states from the last and current time steps. Extensive experiments on diversified sequential modeling tasks are conducted to show the superior generalization ability when there exist changes in the testing environment. Source code is available at \url{https://github.com/ziyuwwang/rig-lstm}.

cs.LG↗
arXiv2023

Object-centric Learning with Cyclic Walks between Parts and Whole

Learning object-centric representations from complex natural environments enables both humans and machines with reasoning abilities from low-level perceptual features. To capture compositional entities of the scene, we proposed cyclic walks between perceptual features extracted from vision transformers and object entities. First, a slot-attention module interfaces with these perceptual features and produces a finite set of slot representations. These slots can bind to any object entities in the scene via inter-slot competitions for attention. Next, we establish entity-feature correspondence with cyclic walks along high transition probability based on the pairwise similarity between perceptual features (aka "parts") and slot-binded object representations (aka "whole"). The whole is greater than its parts and the parts constitute the whole. The part-whole interactions form cycle consistencies, as supervisory signals, to train the slot-attention module. Our rigorous experiments on \textit{seven} image datasets in \textit{three} \textit{unsupervised} tasks demonstrate that the networks trained with our cyclic walks can disentangle foregrounds and backgrounds, discover objects, and segment semantic objects in complex scenes. In contrast to object-centric models attached with a decoder for the pixel-level or feature-level reconstructions, our cyclic walks provide strong learning signals, avoiding computation overheads and enhancing memory efficiency. Our source code and data are available at: \href{https://github.com/ZhangLab-DeepNeuroCogLab/Parts-Whole-Object-Centric-Learning/}{link}.

cs.CV↗
arXiv2023

Secondary proximity effect in a side-coupled double quantum dot structure

Semiconductor quantum dots in close proximity to superconductors may provoke localized bound states within the superconducting energy gap known as Yu-Shiba-Rusinov (YSR) state, which is a promising candidate for constructing Majorana zero modes and topological qubits. Side-coupled double quantum dot systems are ideal platforms revealing the secondary proximity effect. Numerical renormalization group calculations show that, if the central quantum dot can be treated as a noninteracting resonant level, it acts as a superconducting medium due to the ordinary proximity effect. The bound state in the side dot behaves as the case of a single impurity connected to two superconducting leads. The side dot undergoes quantum phase transitions between a singlet state and a doublet state as the Coulomb repulsion, the interdot coupling strength, or the energy level sweeps. Phase diagrams indicate that the phase boundaries could be well illustrated by $Δ\approx c {T_{K2}}$ in all cases, with $Δ$ is the superconducting gap, $T_{K2}$ is the side Kondo temperature and $c$ is of the order $1.0$. These findings offer valuable insights into the secondary proximity effect, and show great importance for designing superconducting quantum devices.

cond-mat.mes-hall↗
arXiv2023

Perfect Coulomb drag and exciton transport in an excitonic insulator

Strongly coupled two-dimensional electron-hole bilayers can give rise to novel quantum Bosonic states: electrons and holes in electrically isolated layers can pair into interlayer excitons, which can form a Bose-Einstein condensate below a critical temperature at zero magnetic field. This state is predicted to feature perfect Coulomb drag, where a current in one layer must be accompanied by an equal but opposite current in the other, and counterflow superconductivity, where the excitons form a superfluid with zero viscosity. Electron-hole bilayers in the strong coupling limit with an excitonic insulator ground state have been recently achieved in semiconducting transition metal dichalcogenide heterostructures, but direct electrical transport measurements remain challenging. Here we use a novel optical spectroscopy to probe the electrical transport of correlated electron-hole fluids in MoSe2/hBN/WSe2 heterostructures. We observe perfect Coulomb drag in the excitonic insulator phase up to a temperature as high as ~15K. Strongly correlated electron and hole transport is also observed at unbalanced electron and hole densities, although the Coulomb drag is not perfect anymore. Meanwhile, the counterflow resistance of interlayer excitons remains finite. These results indicate the formation of an exciton gas in the excitonic insulator which does not condensate into a superfluid at low temperature. Our work also demonstrates that dynamic optical spectroscopy provides a powerful tool for probing novel exciton transport behavior and possible exciton superfluidity in correlated quantum electron-hole fluids.

cond-mat.mes-hall↗
arXiv2023

A volatile polymer stamp for large-scale, etching-free, and ultraclean transfer and assembly of two-dimensional materials and its heterostructures

The intact transfer and assembly of two-dimensional (2D) materials and their heterostructures are critical for their integration into advanced electronic and optical devices. Herein, we report a facile technique called volatile polymer stamping (VPS) to achieve efficient transfer of 2D materials and assembly of large-scale heterojunctions with clean interfaces. The central feature of the VPS technique is the use of volatile polyphthalaldehyde (PPA) together with hydrophobic polystyrene (PS). While PS enables the direct delamination of 2D materials from hydrophilic substrates owing to water intercalation, PPA can protect 2D materials from solution attack and maintain their integrity during PS removal. Thereafter, PPA can be completely removed by thermal annealing at 180 °C. The proposed VPS technique overcomes the limitations of currently used transfer techniques, such as chemical etching during the delamination stage, solution tearing during cleaning, and contamination from polymer residues.

physics.app-ph↗
arXiv2023

Reliable Synthesis of Large-Area Monolayer WS2 Single Crystals, Films, and Heterostructures with Extraordinary Photoluminescence Induced by Water Intercalation

Two-dimensional (2D) transition metal dichalcogenides (TMDs) hold great potential for future low-energy optoelectronics owing to their unique electronic, optical, and mechanical properties. Chemical vapor deposition (CVD) is the technique widely used for the synthesis of large-area TMDs. However, due to high sensitivity to the growth environment, reliable synthesis of monolayer TMDs via CVD remains challenging. Here we develop a controllable CVD process for large-area synthesis of monolayer WS2 crystals, films, and in-plane graphene-WS2 heterostructures by cleaning the reaction tube with hydrochloric acid, sulfuric acid and aqua regia. The concise cleaning process can remove the residual contaminates attached to the CVD reaction tube and crucibles, reducing the nucleation density but enhancing the diffusion length of WS2 species. The photoluminescence (PL) mappings of a WS2 single crystal and film reveal that the extraordinary PL around the edges of a triangular single crystal is induced by ambient water intercalation at the WS2-sapphire interface. The extraordinary PL can be controlled by the choice of substrates with different wettabilities.

cond-mat.mtrl-sci↗
arXiv2023

Neural Residual Radiance Fields for Streamably Free-Viewpoint Videos

The success of the Neural Radiance Fields (NeRFs) for modeling and free-view rendering static objects has inspired numerous attempts on dynamic scenes. Current techniques that utilize neural rendering for facilitating free-view videos (FVVs) are restricted to either offline rendering or are capable of processing only brief sequences with minimal motion. In this paper, we present a novel technique, Residual Radiance Field or ReRF, as a highly compact neural representation to achieve real-time FVV rendering on long-duration dynamic scenes. ReRF explicitly models the residual information between adjacent timestamps in the spatial-temporal feature space, with a global coordinate-based tiny MLP as the feature decoder. Specifically, ReRF employs a compact motion grid along with a residual feature grid to exploit inter-frame feature similarities. We show such a strategy can handle large motions without sacrificing quality. We further present a sequential training scheme to maintain the smoothness and the sparsity of the motion/residual grids. Based on ReRF, we design a special FVV codec that achieves three orders of magnitudes compression rate and provides a companion ReRF player to support online streaming of long-duration FVVs of dynamic scenes. Extensive experiments demonstrate the effectiveness of ReRF for compactly representing dynamic radiance fields, enabling an unprecedented free-viewpoint viewing experience in speed and quality.

cs.CV↗
arXiv2023

Throughput of Hybrid UAV Networks with Scale-Free Topology

Unmanned Aerial Vehicles (UAVs) hold great potential to support a wide range of applications due to the high maneuverability and flexibility. Compared with single UAV, UAV swarm carries out tasks efficiently in harsh environment, where the network resilience is of vital importance to UAV swarm. The network topology has a fundamental impact on the resilience of UAV network. It is discovered that scale-free network topology, as a topology that exists widely in nature, has the ability to enhance the network resilience. Besides, increasing network throughput can enhance the efficiency of information interaction, improving the network resilience. Facing these facts, this paper studies the throughput of UAV Network with scale-free topology. Introducing the hybrid network structure combining both ad hoc transmission mode and cellular transmission mode into UAV Network, the throughput of UAV Network is improved compared with that of pure ad hoc UAV network. Furthermore, this work also investigates the optimal setting of the hop threshold for the selection of ad hoc or cellular transmission mode. It is discovered that the optimal hop threshold is related with the number of UAVs and the parameters of scale-free topology. This paper may motivate the application of hybrid network structure into UAV Network.

cs.IT↗
arXiv2023

HUB: Guiding Learned Optimizers with Continuous Prompt Tuning

Learned optimizers are a crucial component of meta-learning. Recent advancements in scalable learned optimizers have demonstrated their superior performance over hand-designed optimizers in various tasks. However, certain characteristics of these models, such as an unstable learning curve, limited ability to handle unseen tasks and network architectures, difficult-to-control behaviours, and poor performance in fine-tuning tasks impede their widespread adoption. To tackle the issue of generalization in scalable learned optimizers, we propose a hybrid-update-based (HUB) optimization strategy inspired by recent advancements in hard prompt tuning and result selection techniques used in large language and vision models. This approach can be easily applied to any task that involves hand-designed or learned optimizer. By incorporating hand-designed optimizers as the second component in our hybrid approach, we are able to retain the benefits of learned optimizers while stabilizing the training process and, more importantly, improving testing performance. We validate our design through a total of 17 tasks, consisting of thirteen training from scratch and four fine-tuning settings. These tasks vary in model sizes, architectures, or dataset sizes, and the competing optimizers are hyperparameter-tuned. We outperform all competitors in 94% of the tasks with better testing performance. Furthermore, we conduct a theoretical analysis to examine the potential impact of our hybrid strategy on the behaviours and inherited traits of learned optimizers.

cs.LG↗
arXiv2023

PowerGAN: A Machine Learning Approach for Power Side-Channel Attack on Compute-in-Memory Accelerators

Analog compute-in-memory (CIM) systems are promising for deep neural network (DNN) inference acceleration due to their energy efficiency and high throughput. However, as the use of DNNs expands, protecting user input privacy has become increasingly important. In this paper, we identify a potential security vulnerability wherein an adversary can reconstruct the user's private input data from a power side-channel attack, under proper data acquisition and pre-processing, even without knowledge of the DNN model. We further demonstrate a machine learning-based attack approach using a generative adversarial network (GAN) to enhance the data reconstruction. Our results show that the attack methodology is effective in reconstructing user inputs from analog CIM accelerator power leakage, even at large noise levels and after countermeasures are applied. Specifically, we demonstrate the efficacy of our approach on an example of U-Net inference chip for brain tumor detection, and show the original magnetic resonance imaging (MRI) medical images can be successfully reconstructed even at a noise-level of 20% standard deviation of the maximum power signal value. Our study highlights a potential security vulnerability in analog CIM accelerators and raises awareness of using GAN to breach user privacy in such systems.

cs.CR↗
arXiv2023

Bulk-Switching Memristor-based Compute-In-Memory Module for Deep Neural Network Training

The need for deep neural network (DNN) models with higher performance and better functionality leads to the proliferation of very large models. Model training, however, requires intensive computation time and energy. Memristor-based compute-in-memory (CIM) modules can perform vector-matrix multiplication (VMM) in situ and in parallel, and have shown great promises in DNN inference applications. However, CIM-based model training faces challenges due to non-linear weight updates, device variations, and low-precision in analog computing circuits. In this work, we experimentally implement a mixed-precision training scheme to mitigate these effects using a bulk-switching memristor CIM module. Lowprecision CIM modules are used to accelerate the expensive VMM operations, with high precision weight updates accumulated in digital units. Memristor devices are only changed when the accumulated weight update value exceeds a pre-defined threshold. The proposed scheme is implemented with a system-on-chip (SoC) of fully integrated analog CIM modules and digital sub-systems, showing fast convergence of LeNet training to 97.73%. The efficacy of training larger models is evaluated using realistic hardware parameters and shows that that analog CIM modules can enable efficient mix-precision DNN training with accuracy comparable to full-precision software trained models. Additionally, models trained on chip are inherently robust to hardware variations, allowing direct mapping to CIM inference chips without additional re-training.

cs.AR↗
arXiv2023

Side-channel attack analysis on in-memory computing architectures

In-memory computing (IMC) systems have great potential for accelerating data-intensive tasks such as deep neural networks (DNNs). As DNN models are generally highly proprietary, the neural network architectures become valuable targets for attacks. In IMC systems, since the whole model is mapped on chip and weight memory read can be restricted, the pre-mapped DNN model acts as a ``black box'' for users. However, the localized and stationary weight and data patterns may subject IMC systems to other attacks. In this paper, we propose a side-channel attack methodology on IMC architectures. We show that it is possible to extract model architectural information from power trace measurements without any prior knowledge of the neural network. We first developed a simulation framework that can emulate the dynamic power traces of the IMC macros. We then performed side-channel leakage analysis to reverse engineer model information such as the stored layer type, layer sequence, output channel/feature size and convolution kernel size from power traces of the IMC macros. Based on the extracted information, full networks can potentially be reconstructed without any knowledge of the neural network. Finally, we discuss potential countermeasures for building IMC systems that offer resistance to these model extraction attack.

cs.CR↗
arXiv2023

NoiseCAM: Explainable AI for the Boundary Between Noise and Adversarial Attacks

Deep Learning (DL) and Deep Neural Networks (DNNs) are widely used in various domains. However, adversarial attacks can easily mislead a neural network and lead to wrong decisions. Defense mechanisms are highly preferred in safety-critical applications. In this paper, firstly, we use the gradient class activation map (GradCAM) to analyze the behavior deviation of the VGG-16 network when its inputs are mixed with adversarial perturbation or Gaussian noise. In particular, our method can locate vulnerable layers that are sensitive to adversarial perturbation and Gaussian noise. We also show that the behavior deviation of vulnerable layers can be used to detect adversarial examples. Secondly, we propose a novel NoiseCAM algorithm that integrates information from globally and pixel-level weighted class activation maps. Our algorithm is susceptible to adversarial perturbations and will not respond to Gaussian random noise mixed in the inputs. Third, we compare detecting adversarial examples using both behavior deviation and NoiseCAM, and we show that NoiseCAM outperforms behavior deviation modeling in its overall performance. Our work could provide a useful tool to defend against certain adversarial attacks on deep neural networks.

cs.LG↗
arXiv2022

Spectral Representation Learning for Conditional Moment Models

Many problems in causal inference and economics can be formulated in the framework of conditional moment models, which characterize the target function through a collection of conditional moment restrictions. For nonparametric conditional moment models, efficient estimation often relies on preimposed conditions on various measures of ill-posedness of the hypothesis space, which are hard to validate when flexible models are used. In this work, we address this issue by proposing a procedure that automatically learns representations with controlled measures of ill-posedness. Our method approximates a linear representation defined by the spectral decomposition of a conditional expectation operator, which can be used for kernelized estimators and is known to facilitate minimax optimal estimation in certain settings. We show this representation can be efficiently estimated from data, and establish L2 consistency for the resulting estimator. We evaluate the proposed method on proximal causal inference tasks, exhibiting promising performance on high-dimensional, semi-synthetic data.

stat.ML↗
arXiv2022

Application of Deep Q Learning with Simulation Results for Elevator Optimization

This paper presents a methodology for combining programming and mathematics to optimize elevator wait times. Based on simulated user data generated according to the canonical three-peak model of elevator traffic, we first develop a naive model from an intuitive understanding of the logic behind elevators. We take into consideration a general array of features including capacity, acceleration, and maximum wait time thresholds to adequately model realistic circumstances. Using the same evaluation framework, we proceed to develop a Deep Q Learning model in an attempt to match the hard-coded naive approach for elevator control. Throughout the majority of the paper, we work under a Markov Decision Process (MDP) schema, but later explore how the assumption fails to characterize the highly stochastic overall Elevator Group Control System (EGCS).

cs.LG↗