arXiv subjects

Yingyong Qi

Publications and source records attributed to Yingyong Qi.

28 records · Page 2Linked to original sources

arXiv2018

DNQ: Dynamic Network Quantization

Network quantization is an effective method for the deployment of neural networks on memory and energy constrained mobile devices. In this paper, we propose a Dynamic Network Quantization (DNQ) framework which is composed of two modules: a bit-width controller and a quantizer. Unlike most existing quantization methods that use a universal quantization bit-width for the whole network, we utilize policy gradient to train an agent to learn the bit-width of each layer by the bit-width controller. This controller can make a trade-off between accuracy and compression ratio. Given the quantization bit-width sequence, the quantizer adopts the quantization distance as the criterion of the weights importance during quantization. We extensively validate the proposed approach on various main-stream neural networks and obtain impressive results.

cs.LG↗

arXiv2018

BinaryRelax: A Relaxation Approach For Training Deep Neural Networks With Quantized Weights

We propose BinaryRelax, a simple two-phase algorithm, for training deep neural networks with quantized weights. The set constraint that characterizes the quantization of weights is not imposed until the late stage of training, and a sequence of \emph{pseudo} quantized weights is maintained. Specifically, we relax the hard constraint into a continuous regularizer via Moreau envelope, which turns out to be the squared Euclidean distance to the set of quantized weights. The pseudo quantized weights are obtained by linearly interpolating between the float weights and their quantizations. A continuation strategy is adopted to push the weights towards the quantized state by gradually increasing the regularization parameter. In the second phase, exact quantization scheme with a small learning rate is invoked to guarantee fully quantized weights. We test BinaryRelax on the benchmark CIFAR and ImageNet color image datasets to demonstrate the superiority of the relaxed quantization approach and the improved accuracy over the state-of-the-art training methods. Finally, we prove the convergence of BinaryRelax under an approximate orthogonality condition.

cs.CV↗

arXiv2017

Quantization and Training of Low Bit-Width Convolutional Neural Networks for Object Detection

We present LBW-Net, an efficient optimization based method for quantization and training of the low bit-width convolutional neural networks (CNNs). Specifically, we quantize the weights to zero or powers of two by minimizing the Euclidean distance between full-precision weights and quantized weights during backpropagation. We characterize the combinatorial nature of the low bit-width quantization problem. For 2-bit (ternary) CNNs, the quantization of $N$ weights can be done by an exact formula in $O(N\log N)$ complexity. When the bit-width is three and above, we further propose a semi-analytical thresholding scheme with a single free parameter for quantization that is computationally inexpensive. The free parameter is further determined by network retraining and object detection tests. LBW-Net has several desirable advantages over full-precision CNNs, including considerable memory savings, energy efficiency, and faster deployment. Our experiments on PASCAL VOC dataset show that compared with its 32-bit floating-point counterpart, the performance of the 6-bit LBW-Net is nearly lossless in the object detection tasks, and can even do better in some real world visual scenes, while empirically enjoying more than 4$\times$ faster deployment.

cs.LG↗

arXiv2007

A Dynamic Algorithm for Blind Separation of Convolutive Sound Mixtures

We study an efficient dynamic blind source separation algorithm of convolutive sound mixtures based on updating statistical information in the frequency domain, andminimizing the support of time domain demixing filters by a weighted least square method. The permutation and scaling indeterminacies of separation, and concatenations of signals in adjacent time frames are resolved with optimization of $l^1 \times l^\infty$ norm on cross-correlation coefficients at multiple time lags. The algorithm is a direct method without iterations, and is adaptive to the environment. Computations on recorded and synthetic mixtures of speech and music signals show excellent performance.

math.ST↗

arXiv2006

A Many to One Discrete Auditory Transform

A many to one discrete auditory transform is presented to map a sound signal to a perceptually meaningful spectrum on the scale of human auditory filter band widths (critical bands). A generalized inverse is constructed in closed analytical form, preserving the band energy and band signal to noise ratio of the input sound signal. The forward and inverse transforms can be implemented in real time. Experiments on speech and music segments show that the inversion gives a perceptually equivalent though mathematically different sound from the input.

math.NA↗

arXiv2005

An Orthogonal Discrete Auditory Transform

An orthogonal discrete auditory transform (ODAT) from sound signal to spectrum is constructed by combining the auditory spreading matrix of Schroeder et al and the time one map of a discrete nonlocal Schrödinger equation. Thanks to the dispersive smoothing property of the Schrödinger evolution, ODAT spectrum is smoother than that of the discrete Fourier transform (DFT) consistent with human audition. ODAT and DFT are compared in signal denoising tests with spectral thresholding method. The signals are noisy speech segments. ODAT outperforms DFT in signal to noise ratio (SNR) when the noise level is relatively high.

math.NA↗

arXiv2004

An Invertible Discrete Auditory Transform

A discrete auditory transform (DAT) from sound signal to spectrum is presented and shown to be invertible in closed form. The transform preserves energy, and its spectrum is smoother than that of the discrete Fourier transform (DFT) consistent with human audition. DAT and DFT are compared in signal denoising tests with spectral thresholding method. The signals are noisy speech segments. It is found that DAT can gain 5 to 7 decibel (dB) in signal to noise ratio (SNR) over DFT except when the noise level is relatively low.

math.NA↗

arXiv2003

Global well-posedness and multi-tone solutions of a class of nonlinear nonlocal cochlear models in hearing

We study a class of nonlinear nonlocal cochlear models of the transmission line type, describing the motion of basilar membrane (BM) in the cochlea. They are damped dispersive partial differential equations (PDEs) driven by time dependent boundary forcing due to the input sounds. The global well-posedness in time follows from energy estimates. Uniform bounds of solutions hold in case of bounded nonlinear damping. When the input sounds are multi-frequency tones, and the nonlinearity in the PDEs is cubic, we construct smooth quasi-periodic solutions (multi-tone solutions) in the weakly nonlinear regime, where new frequencies are generated due to nonlinear interaction. When the input is two tones at frequencies $f_1$, $f_2$ ($f_1 < f_2$), and high enough intensities, numerical results illustrate the formation of combination tones at $2 f_1 -f_2$ and $2f_2 -f_1$, in agreement with hearing experiments. We visualize the frequency content of solutions through the FFT power spectral density of displacement at selected spatial locations on BM.

nlin.PS↗

arXiv2002

Time Domain Computation of a Nonlinear Nonlocal Cochlear Model with Applications to Multitone Interaction in Hearing

A nonlinear nonlocal cochlear model of the transmission line type is studied in order to capture the multitone interactions and resulting tonal suppression effects. The model can serve as a module for voice signal processing, it is a one dimensional (in space) damped dispersive nonlinear PDE based on mechanics and phenomenology of hearing. It describes the motion of basilar membrane (BM) in the cochlea driven by input pressure waves. Both elastic damping and selective longitudinal fluid damping are present. The former is nonlinear and nonlocal in BM displacement, and plays a key role in capturing tonal interactions. The latter is active only near the exit boundary (helicotrema), and is built in to damp out the remaining long waves. The initial boundary value problem is numerically solved with a semi-implicit second order finite difference method. Solutions reach a multi-frequency quasi-steady state. Numerical results are shown on two tone suppression from both high-frequency and low-frequency sides, consistent with known behavior of two tone suppression. Suppression effects among three tones are demonstrated by showing how the response magnitudes of the fixed two tones are reduced as we vary the third tone in frequency and amplitude. We observe qualitative agreement of our model solutions with existing cat auditory neural data. The model is thus simple and efficient as a processing tool for voice signals.

nlin.PS↗

arXiv2002

Modeling Vocal Fold Motion with a New Hydrodynamic Semi-Continuum Model

Vocal fold (VF) motion is a fundamental process in voice production, and is also a challenging problem for direct numerical computation because the VF dynamics depend on nonlinear coupling of air flow with the response of elastic channels (VF), which undergo opening and closing, and induce internal flow separation. A traditional modeling approach makes use of steady flow approximation or Bernoulli's law which is known to be invalid during VF opening. We present a new hydrodynamic semi-continuum system for VF motion. The airflow is modeled by a quasi-one dimensional continuum aerodynamic system, and the VF by a classical lumped two mass system. The reduced flow system contains the Bernoulli's law as a special case, and is derivable from the two dimensional compressible Navier-Stokes equations. Since we do not make steady flow approximation, we are able to capture transients and rapid changes of solutions, e.g. the double pressure peaks at opening and closing stages of VF motion consistent with experimental data. We demonstrate numerically that our system is robust, and models in-vivo VF oscillation more physically. It is also much simpler than a full two-dimensional Navier-Stokes system.

nlin.PS↗