arXiv subjects

Yueming Wang

Publications and source records attributed to Yueming Wang.

33 records · Page 2Linked to original sources

arXiv2023

MindGPT: Interpreting What You See with Non-invasive Brain Recordings

Decoding of seen visual contents with non-invasive brain recordings has important scientific and practical values. Efforts have been made to recover the seen images from brain signals. However, most existing approaches cannot faithfully reflect the visual contents due to insufficient image quality or semantic mismatches. Compared with reconstructing pixel-level visual images, speaking is a more efficient and effective way to explain visual information. Here we introduce a non-invasive neural decoder, termed as MindGPT, which interprets perceived visual stimuli into natural languages from fMRI signals. Specifically, our model builds upon a visually guided neural encoder with a cross-attention mechanism, which permits us to guide latent neural representations towards a desired language semantic direction in an end-to-end manner by the collaborative use of the large language model GPT. By doing so, we found that the neural representations of the MindGPT are explainable, which can be used to evaluate the contributions of visual properties to language semantics. Our experiments show that the generated word sequences truthfully represented the visual information (with essential details) conveyed in the seen stimuli. The results also suggested that with respect to language decoding tasks, the higher visual cortex (HVC) is more semantically informative than the lower visual cortex (LVC), and using only the HVC can recover most of the semantic information. The code of the MindGPT model will be publicly available at https://github.com/JxuanC/MindGPT.

cs.CV↗

arXiv2023

A Human-Machine Joint Learning Framework to Boost Endogenous BCI Training

Brain-computer interfaces (BCIs) provide a direct pathway from the brain to external devices and have demonstrated great potential for assistive and rehabilitation technologies. Endogenous BCIs based on electroencephalogram (EEG) signals, such as motor imagery (MI) BCIs, can provide some level of control. However, mastering spontaneous BCI control requires the users to generate discriminative and stable brain signal patterns by imagery, which is challenging and is usually achieved over a very long training time (weeks/months). Here, we propose a human-machine joint learning framework to boost the learning process in endogenous BCIs, by guiding the user to generate brain signals towards an optimal distribution estimated by the decoder, given the historical brain signals of the user. To this end, we firstly model the human-machine joint learning process in a uniform formulation. Then a human-machine joint learning framework is proposed: 1) for the human side, we model the learning process in a sequential trial-and-error scenario and propose a novel ``copy/new'' feedback paradigm to help shape the signal generation of the subject toward the optimal distribution; 2) for the machine side, we propose a novel adaptive learning algorithm to learn an optimal signal distribution along with the subject's learning process. Specifically, the decoder reweighs the brain signals generated by the subject to focus more on ``good'' samples to cope with the learning process of the subject. Online and psuedo-online BCI experiments with 18 healthy subjects demonstrated the advantages of the proposed joint learning process over co-adaptive approaches in both learning efficiency and effectiveness.

cs.HC↗

arXiv2023

ESL-SNNs: An Evolutionary Structure Learning Strategy for Spiking Neural Networks

Spiking neural networks (SNNs) have manifested remarkable advantages in power consumption and event-driven property during the inference process. To take full advantage of low power consumption and improve the efficiency of these models further, the pruning methods have been explored to find sparse SNNs without redundancy connections after training. However, parameter redundancy still hinders the efficiency of SNNs during training. In the human brain, the rewiring process of neural networks is highly dynamic, while synaptic connections maintain relatively sparse during brain development. Inspired by this, here we propose an efficient evolutionary structure learning (ESL) framework for SNNs, named ESL-SNNs, to implement the sparse SNN training from scratch. The pruning and regeneration of synaptic connections in SNNs evolve dynamically during learning, yet keep the structural sparsity at a certain level. As a result, the ESL-SNNs can search for optimal sparse connectivity by exploring all possible parameters across time. Our experiments show that the proposed ESL-SNNs framework is able to learn SNNs with sparse structures effectively while reducing the limited accuracy. The ESL-SNNs achieve merely 0.28% accuracy loss with 10% connection density on the DVS-Cifar10 dataset. Our work presents a brand-new approach for sparse training of SNNs from scratch with biologically plausible evolutionary mechanisms, closing the gap in the expressibility between sparse training and dense training. Hence, it has great potential for SNN lightweight training and inference with low power consumption and small memory usage.

cs.NE↗

arXiv2023

Revisiting Neural Retrieval on Accelerators

Retrieval finds a small number of relevant candidates from a large corpus for information retrieval and recommendation applications. A key component of retrieval is to model (user, item) similarity, which is commonly represented as the dot product of two learned embeddings. This formulation permits efficient inference, commonly known as Maximum Inner Product Search (MIPS). Despite its popularity, dot products cannot capture complex user-item interactions, which are multifaceted and likely high rank. We hence examine non-dot-product retrieval settings on accelerators, and propose \textit{mixture of logits} (MoL), which models (user, item) similarity as an adaptive composition of elementary similarity functions. This new formulation is expressive, capable of modeling high rank (user, item) interactions, and further generalizes to the long tail. When combined with a hierarchical retrieval strategy, \textit{h-indexer}, we are able to scale up MoL to 100M corpus on a single GPU with latency comparable to MIPS baselines. On public datasets, our approach leads to uplifts of up to 77.3\% in hit rate (HR). Experiments on a large recommendation surface at Meta showed strong metric gains and reduced popularity bias, validating the proposed approach's performance and improved generalization.

cs.LG↗

arXiv2023

Decoding Chinese phonemes from intracortical brain signals with hyperbolic-space neural representations

Speech brain-computer interfaces (BCIs), which translate brain signals into spoken words or sentences, have shown significant potential for high-performance BCI communication. Phonemes are the fundamental units of pronunciation in most languages. While existing speech BCIs have largely focused on English, where words contain diverse compositions of phonemes, Chinese Mandarin is a monosyllabic language, with words typically consisting of a consonant and a vowel. This feature makes it feasible to develop high-performance Mandarin speech BCIs by decoding phonemes directly from neural signals. This study aimed to decode spoken Mandarin phonemes using intracortical neural signals. We observed that phonemes with similar pronunciations were often represented by inseparable neural patterns, leading to confusion in phoneme decoding. This finding suggests that the neural representation of spoken phonemes has a hierarchical structure. To account for this, we proposed learning the neural representation of phoneme pronunciation in a hyperbolic space, where the hierarchical structure could be more naturally optimized. Experiments with intracortical neural signals from a Chinese participant showed that the proposed model learned discriminative and interpretable hierarchical phoneme representations from neural signals, significantly improving Chinese phoneme decoding performance and achieving state-of-the-art. The findings demonstrate the feasibility of constructing high-performance Chinese speech BCIs based on phoneme decoding.

cs.HC↗

arXiv2023

LaSNN: Layer-wise ANN-to-SNN Distillation for Effective and Efficient Training in Deep Spiking Neural Networks

Spiking Neural Networks (SNNs) are biologically realistic and practically promising in low-power computation because of their event-driven mechanism. Usually, the training of SNNs suffers accuracy loss on various tasks, yielding an inferior performance compared with ANNs. A conversion scheme is proposed to obtain competitive accuracy by mapping trained ANNs' parameters to SNNs with the same structures. However, an enormous number of time steps are required for these converted SNNs, thus losing the energy-efficient benefit. Utilizing both the accuracy advantages of ANNs and the computing efficiency of SNNs, a novel SNN training framework is proposed, namely layer-wise ANN-to-SNN knowledge distillation (LaSNN). In order to achieve competitive accuracy and reduced inference latency, LaSNN transfers the learning from a well-trained ANN to a small SNN by distilling the knowledge other than converting the parameters of ANN. The information gap between heterogeneous ANN and SNN is bridged by introducing the attention scheme, the knowledge in an ANN is effectively compressed and then efficiently transferred by utilizing our layer-wise distillation paradigm. We conduct detailed experiments to demonstrate the effectiveness, efficacy, and scalability of LaSNN on three benchmark data sets (CIFAR-10, CIFAR-100, and Tiny ImageNet). We achieve competitive top-1 accuracy compared to ANNs and 20x faster inference than converted SNNs with similar performance. More importantly, LaSNN is dexterous and extensible that can be effortlessly developed for SNNs with different architectures/depths and input encoding methods, contributing to their potential development.

cs.NE↗

arXiv2023

How Charge Carrier Exchange between Absorber and Contact influences Time Constants in the Frequency Domain Response of Perovskite Solar Cells

A model is derived for the frequency- and time-domain opto-electronic response of perovskite solar cells (PSCs) that emphasizes the role of charge carrier exchange, .i.e. extraction and injection, from (to) the perovskite through the transport layer to (from) the collecting electrode. This process is described by a charge carrier exchange velocity that depends on the mobility and electric field inside the transport layer. The losses implied by this process are modelled in an equivalent circuit model in the form of a voltage-dependent transport layer resistance. The analysis of the model predicts that the voltage dependence of the measured time constants allows discriminating situations where the transport layer properties dominate the experimental response. Application of this method to experimental impedance spectroscopy data identifies charge extraction velocities between 1-100 cm/s at 1 sun open-circuit conditions for p-i-n PSCs with PTAA as the hole transport layer, that corresponds to transport layer mobilities between 10^-4 - 3 x 10^-3 cm^2V^-1s^-1. The model paves the way for accurate estimation of photocurrent and fill factor losses in PSCs caused by the low mobilities in the transport layers, using small perturbation measurements in the time and frequency domain.

cond-mat.mtrl-sci↗

arXiv2022

Dynamic Ensemble Bayesian Filter for Robust Control of a Human Brain-machine Interface

Objective: Brain-machine interfaces (BMIs) aim to provide direct brain control of devices such as prostheses and computer cursors, which have demonstrated great potential for mobility restoration. One major limitation of current BMIs lies in the unstable performance in online control due to the variability of neural signals, which seriously hinders the clinical availability of BMIs. Method: To deal with the neural variability in online BMI control, we propose a dynamic ensemble Bayesian filter (DyEnsemble). DyEnsemble extends Bayesian filters with a dynamic measurement model, which adjusts its parameters in time adaptively with neural changes. This is achieved by learning a pool of candidate functions and dynamically weighting and assembling them according to neural signals. In this way, DyEnsemble copes with variability in signals and improves the robustness of online control. Results: Online BMI experiments with a human participant demonstrate that, compared with the velocity Kalman filter, DyEnsemble significantly improves the control accuracy (increases the success rate by 13.9% and reduces the reach time by 13.5% in the random target pursuit task) and robustness (performs more stably over different experiment days). Conclusion: Our results demonstrate the superiority of DyEnsemble in online BMI control. Significance: DyEnsemble frames a novel and flexible framework for robust neural decoding, which is beneficial to different neural decoding applications.

cs.LG↗

arXiv2021

Layer-dependent Optical and Dielectric Properties of Large-size PdSe$_2$ Films Grown by Chemical Vapor Deposition

Palladium diselenide (PdSe$_2$), a new type of two-dimensional noble metal dihalides (NMDCs), has received widespread attention for its excellent electrical and optoelectronic properties. Herein, high-quality continuous centimeter-scale PdSe$_2$ films with layers in the range of 3L-15L were grown using Chemical Vapor Deposition (CVD) method. The absorption spectra and DFT calculations revealed that the bandgap of the PdSe$_2$ films decreased with increasing number of layers, which is due to PdSe$_2$ enhancement of orbital hybridization. Spectroscopic ellipsometry (SE) analysis shows that PdSe2 has significant layer-dependent optical and dielectric properties. This is mainly due to the unique strong exciton effect of the thin PdSe$_2$ film in the UV band. In particular, the effect of temperature on the optical properties of PdSe$_2$ films was also observed, and the thermo-optical coefficients of PdSe$_2$ films with different number of layers were calculated. This study provides fundamental guidance for the fabrication and optimization of PdSe$_2$-based optoelectronic devices.

cond-mat.mtrl-sci↗

arXiv2019

Dynamic Ensemble Modeling Approach to Nonstationary Neural Decoding in Brain-Computer Interfaces

Brain-computer interfaces (BCIs) have enabled prosthetic device control by decoding motor movements from neural activities. Neural signals recorded from cortex exhibit nonstationary property due to abrupt noises and neuroplastic changes in brain activities during motor control. Current state-of-the-art neural signal decoders such as Kalman filter assume fixed relationship between neural activities and motor movements, thus will fail if this assumption is not satisfied. We propose a dynamic ensemble modeling (DyEnsemble) approach that is capable of adapting to changes in neural signals by employing a proper combination of decoding functions. The DyEnsemble method firstly learns a set of diverse candidate models. Then, it dynamically selects and combines these models online according to Bayesian updating mechanism. Our method can mitigate the effect of noises and cope with different task behaviors by automatic model switching, thus gives more accurate predictions. Experiments with neural data demonstrate that the DyEnsemble method outperforms Kalman filters remarkably, and its advantage is more obvious with noisy signals.

eess.SP↗

arXiv2019

Quantum solitons in spin-orbit-coupled Bose-Bose mixtures

Recent experimental and theoretical results show that weakly interacting atomic Bose-Bose mixtures with attractive interspecies interaction are stabilized by beyond-mean-field effects. Here we consider the peculiar properties of these systems in a strictly one-dimensional configuration, taking also into account the nontrivial role of spin-orbit and Rabi couplings. We show that when the value of inter- and intraspecies interaction strengths are such that mean-field contributions to the energy cancel, a self-bound bright soliton fully governed by quantum fluctuations exists. We derive the phase diagram of the phase transition between a single-peak soliton and a multipeak (striped) soliton, produced by the interplay between spin-orbit, Rabi couplings and beyond-mean-field effects, which also affect the breathing mode frequency of the atomic cloud. Finally, we prove that a phase imprinting of the single-peak soliton leads to a self-confined propagating solitary wave even in the presence of spin-orbit coupling.

cond-mat.quant-gas↗

arXiv2017

Spectral Sparse Representation for Clustering: Evolved from PCA, K-means, Laplacian Eigenmap, and Ratio Cut

Dimensionality reduction, cluster analysis, and sparse representation are basic components in machine learning. However, their relationships have not yet been fully investigated. In this paper, we find that the spectral graph theory underlies a series of these elementary methods and can unify them into a complete framework. The methods include PCA, K-means, Laplacian eigenmap (LE), ratio cut (Rcut), and a new sparse representation method developed by us, called spectral sparse representation (SSR). Further, extended relations to conventional over-complete sparse representations (e.g., method of optimal directions, KSVD), manifold learning (e.g., kernel PCA, multidimensional scaling, Isomap, locally linear embedding), and subspace clustering (e.g., sparse subspace clustering, low-rank representation) are incorporated. We show that, under an ideal condition from the spectral graph theory, PCA, K-means, LE, and Rcut are unified together. And when the condition is relaxed, the unification evolves to SSR, which lies in the intermediate between PCA/LE and K-mean/Rcut. An efficient algorithm, NSCrt, is developed to solve the sparse codes of SSR. SSR combines merits of both sides: its sparse codes reduce dimensionality of data meanwhile revealing cluster structure. For its inherent relation to cluster analysis, the codes of SSR can be directly used for clustering. Scut, a clustering approach derived from SSR reaches the state-of-the-art performance in the spectral clustering family. The one-shot solution obtained by Scut is comparable to the optimal result of K-means that are run many times. Experiments on various data sets demonstrate the properties and strengths of SSR, NSCrt, and Scut.

cs.CV↗

arXiv2016

Directed tunneling of a prescribed number of dipolar bosons in shaken triple-well potentials

We propose a scheme for precise control of tunneling dynamics of dipolar bosons in shaken triple-well potentials. In the high-frequency regimes and under the resonance conditions, we have analytically and numerically demonstrated that we can transport a priori prescribed number of dipolar bosons along different pathways and different directions by adjusting the driving parameters. These results extend the previous many-body selective coherent destruction of tunneling (CDT) schemes for nondipolar bosons in double-well potentials[Phys. Rev. Lett. 103, 133002 (2009); Phys. Rev. A 86, 044102 (2012)], thus offering an efficient way to design the long-range coherent quantum transportation.

quant-ph↗

arXiv2015

Robust Face Recognition by Constrained Part-based Alignment

Developing a reliable and practical face recognition system is a long-standing goal in computer vision research. Existing literature suggests that pixel-wise face alignment is the key to achieve high-accuracy face recognition. By assuming a human face as piece-wise planar surfaces, where each surface corresponds to a facial part, we develop in this paper a Constrained Part-based Alignment (CPA) algorithm for face recognition across pose and/or expression. Our proposed algorithm is based on a trainable CPA model, which learns appearance evidence of individual parts and a tree-structured shape configuration among different parts. Given a probe face, CPA simultaneously aligns all its parts by fitting them to the appearance evidence with consideration of the constraint from the tree-structured shape configuration. This objective is formulated as a norm minimization problem regularized by graph likelihoods. CPA can be easily integrated with many existing classifiers to perform part-based face recognition. Extensive experiments on benchmark face datasets show that CPA outperforms or is on par with existing methods for robust face recognition across pose, expression, and/or illumination changes.

cs.CV↗

arXiv2014

Sparse Principal Component Analysis via Rotation and Truncation

Sparse principal component analysis (sparse PCA) aims at finding a sparse basis to improve the interpretability over the dense basis of PCA, meanwhile the sparse basis should cover the data subspace as much as possible. In contrast to most of existing work which deal with the problem by adding some sparsity penalties on various objectives of PCA, in this paper, we propose a new method SPCArt, whose motivation is to find a rotation matrix and a sparse basis such that the sparse basis approximates the basis of PCA after the rotation. The algorithm of SPCArt consists of three alternating steps: rotate PCA basis, truncate small entries, and update the rotation matrix. Its performance bounds are also given. SPCArt is efficient, with each iteration scaling linearly with the data dimension. It is easy to choose parameters in SPCArt, due to its explicit physical explanations. Besides, we give a unified view to several existing sparse PCA methods and discuss the connection with SPCArt. Some ideas in SPCArt are extended to GPower, a popular sparse PCA algorithm, to overcome its drawback. Experimental results demonstrate that SPCArt achieves the state-of-the-art performance. It also achieves a good tradeoff among various criteria, including sparsity, explained variance, orthogonality, balance of sparsity among loadings, and computational speed.

cs.LG↗