The high efficiency in computation and storage makes hashing (including binary quantization) a common strategy large-scale retrieval systems. To alleviate the reliance on expensive annotations, unsupervised deep becomes an important research problem. This paper provides novel solution to quantization, namely Contrastive Quantization with Code Memory (MeCoQ). Different from existing reconstructi...

This thesis is an investigation of robust vector quantization, with the purpose of providing a system for the application of data compression and speech enhancement. Vector quantization is widely used in data compression systems. However, the performance of these systems will degrade in a noisy environment. The proposed robust vector quantization system solves the problem of optimal quantizatio...

This paper proposes new method for designing scalar quantizers. The proposed method, denoted as hybrid method, combines two quantization techniques, the companding technique and the Lloyd-Max's algorithm. In this paper an exact and complete analysis of the hybrid quantization method considering the Laplacian input signals is carried out. Furthermore, two approaches to the problem of finding the...

We propose a multiscale quantization approach for fast similarity search on large, high-dimensional datasets. The key insight of the approach is that quantization methods, in particular product quantization, perform poorly when there is large variance in the norms of the data points. This is a common scenario for realworld datasets, especially when doing product quantization of residuals obtain...

The effects of sampling and quantization on frequency estimation for a single sinusoid are investigated. Cramér–Rao bound for 1-bit quantization is derived and compared with the limit of infinite quantization. It is found that 1-bit quantization gives a slightly worse performance, however, with a dramatic increase of variance at certain frequencies. This can be avoided by using four times overs...

We compare three approaches to the quantization of (2+1)dimensional gravity with a negative cosmological constant: reduced phase space quantization with the York time slicing, quantization of the algebra of holonomies, and quantization of the space of classical solutions. The relationships among these quantum theories allow us to define and interpret time-dependent operators in the “frozen time...

We develop a two stage split vector quantization method with optimum bit allocation, for achieving minimum computational complexity. This also results in much lower memory requirement than the recently proposed switched split vector quantization method. To improve the rate-distortion performance further, a region specific normalization is introduced, which results in 1 bit/vector improvement ov...

We consider quantization along null-surfaces. We note that null-surfaces are characteristic surfaces, which requires a diierent treatment than the usual equal-time Cauchy initial data quantization. We examine this for quantum eld theories and show that by considering quantization on null-surfaces as a characterisitc initial value problem, we can avoid initial relations-constraints-between the e...

This note is an overview of the Poisson sigma model (PSM) and its applications in deformation quantization. Reduction of coisotropic and prePoisson submanifolds, their appearance as branes of the PSM, quantization in terms of L∞and A∞-algebras, and bimodule structures are recalled. As an application, an “almost” functorial quantization of Poisson maps is presented if no anomalies occur. This le...

This paper describes a technique to compress images based on vector quantization. The vector quantizer is designed to reduce both perceptual irrelevancy and mathematical redundancy. This is done without using transforms and entropic coding, which are normally used respectively prior and after quantization. Because of its structure, the vector quantizer can be implemented efficiently as a unifor...