Holographic Associative Memory Research Articles

Experiments with holographic associative memory

Abstract We reiterate the theoretical basics of holographic associative memory, and conduct two experiments. During the first experiment, we teach the system many associations, while during the second experiment, we teach it only one association. In both cases, the recalling capability of the system is examined from different aspects.

Application of the associative recovery method in the challenges of increase informativity of distorted images and detection of minor changes in stored samples

The object of research is optical-electronic methods and related digital information processing. One of the most problematic areas is the reconstruction of missing parts of the stored data and the inability to detect minor changes in the stored samples, as well as the reconstruction of the entire corrected template from its incomplete version. As part of the study, a correlation-optical approach to the problem of holographic associative memory was used, which made it possible to achieve highly efficient heteroassociative reconstruction of the entire corrected template from its incomplete version. The analysis of hologram models with phantom images and nonlinearly recorded holograms read in the associative mode shows a wide range of useful possibilities. It is primed not only in the tasks of reconstruction of data, but also in the case of insignificant changes in savings, highly effective heteroassociative reconstruction based on a non-interference mechanism. The analysis of the results of the correlation-optical approach to the problem of holographic associative memory shows that the described method opens up additional opportunities for solving the problems of detecting small changes in the object scene, which is important, in particular, for early registration of events and phenomena. It is related to the fact that the detection and localization of changes is carried out according to the difference in intensity across the image field (the effect of brightness inversion in the phantom image of referenceless hologram): the brightness of the image of the changed area is higher, and to a greater extent, the smaller the changes compared to the reference image. It should be especially noted that the specified properties of the nonlinear-holographic associative memory are realized not algorithmically, but physically, taking into account the fundamental nonlinearity of all natural processes, which is neglected when conducting a superficial (in the first approximation) analysis. Physical modeling of associative memory based on second-order holograms does not involve any circuit complications compared to the standard holographic procedure.

HOLOGRAPHIC ASSOCIATIVE MEMORY PROBLEM WITHIN CORRELATION-OPTICAL APPROACH

Проблема голографічної асоціативної пам’яті обговорюється в рамках кореляційно-оптичного підходу. Аналіз моделей безопорної голограми та нелінійно зареєстрованих позаосевих голограм, зчитуваних в асоціативному режимі, показує широкий спектр корисних властивостей таких голограм не лише для відновлення відсутньої частини збереженого масиву даних, а й для детектування малих змін (помилок) у записаній картині, для відновлення повного масиву даних з корекцією помилок за неповною версією масиву, а також для здійснення високоефективного гетероасоціативного відновлення на базі неінтерференційного механізму зв’язку парціальних сигналів. Показано, що візуально оцінювана якість відновленого асоціативного відгуку може наближатись до якості зображення, що досягається у звичайній позаосевій голографії.

Self-conjugation heteroassociative memories using thin static nonlinearly recorded holograms

The problem of all-optical holographic associative memory is elaborated with account of recording nonlinearity of thin static off-axis holograms. It is shown that the recording nonlinearity results in reconstruction of angularly separated diffraction orders, in one of which (minus 2nd) the complex congugate associative response is reconstructed as such a hologram is read out by incomplete version of the stored pattern. Being angularly separated from the readout beam, the phase conjugate response occurs be error-corrected. Moreover, recording nonlinearity provides unique feasibilities for associative coupling and reconstruction of heteroassociations beteween complex patternts accumulated at one carrier without interference. This feasibility obliged just to natural recording nonlinearity (`nonidealness') of registration medium leads to limit generalization of the concept of holographic associative memory.

Optical correlation approach to all-optical holographic associative memory

The problem of all-optical holographic associative memory is discussed within the framework of optical correlation approach. Analysis of the models of ghost-image holograms and nonlenearly recorded off-axis holograms read out in associative regime shows a wide set of useful abilities of such holograms to not only reconstruct the missing part of the stored data, but also detect small changes ("errors") in the stored pattern, reconstruct a whole error-corrected pattern from its incomplete version, as well as perform high-efficient heteroassociative reconstruction based on non-interference mechanism of coupling of partial patterns. It is shown that visually estimated quality of reconstructed associative response can approach quality of reconstruction provided by common off-axis hologram. We time this overview to the 50th anniversary of the holographic associative memory.

Holographic memory-based Bayesian optimization algorithm (HM-BOA) in dynamic environments

This paper presents a new evolutionary dynamic optimization algorithm, holographic memory-based Bayesian optimization algorithm (HM-BOA), whose objective is to address the weaknesses of sequential memorybased dynamic optimization approaches. To this end, holographic associative neural memory is applied to one of the recent successful memory-based evolutionary methods, DBN-MBOA (memory-based BOA with dynamic Bayesian networks). Holographic memory is appropriate for encoding environmental changes since its stimulus and response data are represented by a vector of complex numbers such that the phase and the magnitude denote the information and its confidence level, respectively. In the learning process in HM-BOA, holographic memory is trained by probabilistic models at every environmental change. Its weight matrix contains abstract information obtained from previous changes and is used for constructing a new probabilistic model when the environment changes. The unique features of HM-BOA are: 1) the stored information can be generalized, and 2) a small amount of memory is required for storing the probabilistic models. Experimental results adduce grounds for its effectiveness especially in random environments.

Holographic associative memory by phase conjugate of four-wave-mixing in Sc:Fe:LiNbO3 crystal

Sc:Fe:LiNbO3 crystal series doped with Sc2O3 in different levels have been grown by Czochralski technology. The optical damage resistance ability of crystals improves significantly with the increase of Sc doping amount. The intensity threshold of Sc(3 mol%):Fe:LiNbO3 is two orders of magnitude higher than that of Fe:LiNbO3. By four-wave-mixing method, the phase conjugate refractivity and response time are determined. The holographic associative memory system has been demonstrated by using Sc(3 mol%):Fe:LiNbO3 as recording material and Sc(2 mol%):Fe:LiNbO3 as phase conjugate mirror. The reconstructed image is complete and clear.

A holographic memory approach for pollution forecasting in a high-density urban environment

In this work, the Holographic Associative Memory (HAM) paradigm was used as the core of a forecasting software tool for benzopyrene estimations near a highly populated zone. The presented tool was trained with data coming from a monitoring station near a steel plant in Genova, Italy. The decoding of test stimuli was performed with two different methods, the holographic complex number technique (HCD) and the closest holographic neighbor decoding (CHN). The cost–performance relation of both methods is outlined and compared. The atmospheric scenarios used for modeling benzopyrene behavior contained meteorological and chemical variables correlated to the formation and dispersion of such contaminant. The obtained results show an accurate performance of the HAM method either for identifying the main features involved in benzopyrene estimation and for the forecasting itself. Finally, some concluding remarks regarding the performance of both decoding methods are presented.

Invariant performance of a volume holographic wavelet correlation processor

A multichannel optical correlation processor based on volume holographic associative memory in a photorefractive crystal and wavelet transform is proposed for human face recognition. Distortions due to shift, rotation, scale, and partial hiding are studied to understand invariant performance of the processor. Our results show that shift-invariance and rotation-invariance are key problems for practical applications of the processor to human face recognition. Theoretical analysis and simulation conclude that the focal length of the Fourier transform lens is the main factor to affect shift-invariance of the processor. Shift-invariance would be improved if the focal length were enlarged. With regard to rotation-invariance, a novel mechanism to recognize human faces with any rotation angle is proposed and testified by experiments. The processor is more practical with the improvement of invariance.

Multiplexing of volume holographic wavelet correlation processor

Volume holographic associative memory in a photorefractive crystal provides an inherent mechanism to develop a multi-channel correlation identification system with high parallelism. Wavelet transform is introduced to improve discrimination of the system. We first investigate parameters of the system for parallelism enhancement, and then study multiplexing of the system on input objects and wavelet filters. A general volume holographic wavelet correlation processor has a single input-object channel and a single wavelet-filtering channel. In other words, it can only process one input object with one wavelet filter at a same time. Based on the fact that a volume holographic correlator is not a shift-invariant system, multiplexing of input objects is proposed to improve parallelism of the processor. As a result, several input objects can be recognized simultaneously. Multiplexing of wavelet filters with different wavelet parameters is also achieved by a Dammann grating. Wavelet correlation outputs with different filters are synthesized to improve recognition accuracy of the processor. Corresponding experimental results in human face recognition are given. The combination of the input object multiplexing and the wavelet filter multiplexing is also described.

Biomolecular Electronics: Protein-Based Associative Processors and Volumetric Memories

The promise of new architectures and more cost-effective miniaturization has prompted interest in molecular and biomolecular electronics. Bioelectronics offers valuable near-term potential, because evolution and natural selection have optimized many biological molecules to perform tasks that are required for device applications. The light-transducing protein bacteriorhodopsin provides not only an efficient photonic material, but also a versatile template for device creation and optimization via both chemical modification and genetic engineering. We examine here the use of this protein as the active component in holographic associative memories as well as branched-photocycle three-dimensional optical memories. The associative memory is based on a Fourier transform optical loop and utilizes the real-time holographic properties of the protein thin films. The three-dimensional memory utilizes an unusual branching reaction that creates a long-lived photoproduct. By using a sequential multiphoton process, parallel write, read, and erase processes can be carried out without disturbing data outside of the doubly irradiated volume elements. The methods and procedures of prototyping these bioelectronic devices are discussed. We also examine current efforts to optimize the protein memory medium by using chemical and genetic methods.

Handwritten digit recognition by means of a holographic associative memory

In this paper, a holographic associative memory (HAM) is proposed for recognizing handwritten variations of the ten digits. First, the handwritten characters were taken from the NIST standard database in order to extract relevant features from each one of them. Each digit was thus represented as a vector of 112 features constructed by dividing each character into 16 equal-sized partitions, each one used to extract seven different features for recognition. Second, these feature vectors, and reduced combinations of them, were input to train several HAM systems respectively. Then, all these memories were tested with a new set of patterns and the lowest-error HAM was chosen as the best training set. The features used in this last memory were taken as the most significant variables for describing each digit in the database. Finally, these most significant features were used to show the behaviour of the recognition rate when training the HAM with reduced training sets. Some final conclusions are reported and future work directions are proposed.

Characteristics of multidimensional holographic associative memory in retrieval with dynamically localizable attention

This paper presents the performance analysis (capacity and retrieval accuracy) of multidimensional holographic associative memory (MHAC). MHAC has the unique ability to retrieve pattern-associations with changeable attention. In attention actuated retrieval the user can dynamically select any subset of the elements in the example query pattern and expect the memory to confine its associative match only within the specified field of attention. Existing artificial associative memories lack this ability. Also most of these models need at least 50% of bits in the input pattern to be correct for successful retrieval. MHAC, with the unique ability of localizable attention, can retrieve information correctly even with cues as small as 10% of the query frame. This paper investigates the performance of MHAC in attention actuated retrieval both analytically and experimentally. Besides confirmation, the experiments also identify an operational range space (ORS) for this memory within which various attention based applications can be built with a performance guarantee.

Application of wavelet-matched filters to holographic associative memories

We propose a novel method for improving association abilities of a holographic associative memory by using wavelet-matched filtering. Based on an edge-enhancement and multiscaling properties of the wavelet transform,a sharp correlation peak between a partial input and a stored datum can be obtained. This leads to higher associations abilities. Auto- and hetero-association operations are demonstrated experimentally.

Implementation of real-time holographic associative memory using spatial-encoding

A photorefractive optical processor for accurate holographic associative memory that uses a spatial-encoding method is presented. The associative memory of both binary images and gray-scale images can be performed with this processor. The experimental results have proved that spatial-encoding can efficiently increase the addressing accuracy of real-time holographic associative memory in a simple way.

Implementation of real-time holographic associative memory using spatial-encoding

Implementation of real-time holographic associative memory using spatial-encoding

Holographic associative memory using a coherently induced double phase conjugate mirror

We demonstrate a new method to implement holographic associative memory based on a coherently induced double phase conjugate mirror that has phase-conjugate reflectivity larger than unity for weak reference signal over a large range of pumping-reference intensity ratio. The phase conjugator, operated with a copper-doped KNSBN crystal, serves as a novel nonlinear feedback and gain element for the associative memory system. A vanadium-doped KNSBN crystal as storage medium improves the response speed of the holographic recording process. A stored image has been completely retrieved when a partial input addresses the associative memory system.

Overview of hybrid optical neural networks

This paper reviews optical architectures for the implementation of hybrid neural networks. Optics is mainly applied to implementing the matrix-vector or tensor-matrix multiplication. In addition, the general background of neural networks as well as a brief discussion on holographic associative memory are also given.

Nearly optimal correlations for shift-invariant associative memories

The optical implementation of the Hopfield algorithm in shift-invariant holographic associative memories is based on the use of correlators with matched filters. However, it is well known that such correlators have poor discrimination. We propose nearly optimal correlation designs for associative memories based on correlation filters that have maximum discrimination ability. These new designs avoid large cross-correlation-peak terms caused by a mismatch between partial input and the fully stored information in the filter. These solutions rely on whitened spectra of the stored and the recalled information.Computer simulations are made of eight different combinations.

Holographic associative memory with accurate addressing

We demonstrate the weakness of the real-time holographic associative memory implementation in addressing accuracy and describe a novel method to implement associative memory with accurate addressing. In this method, the memory images and their background images are stored in two common holographic memory systems, and the partial image and its background image are simultaneously used to address the two holographic memory systems. The experimental results have proved the system feasible. In addition, we describe a method of associative memory with accurate addressing in a common real-time holographic memory system using combinatorial memory images and combinatorial addressing images.