A CAM-Based Information Detection Hardware System for Fast Image Matching on FPGA

Le, Duc-Hung; Cao, Tran-Bao-Thuong; Inoue, Katsumi; Pham, Cong‐Kha

doi:10.1587/transele.e97.c.65

Cited by 12 publications

(16 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If there is another method to mimic highly and efficiently the human brain that does not stick to arithmetic calculation, it will be one of the next important AI technology. Set operating processor (SOP) is a technology invented in view of the above problem [2]- [12].…”

Section: Present Situations and Problems Of Aimentioning

confidence: 99%

See 1 more Smart Citation

The Memorism Processor: Towards a Memory-Based Artificially Intelligence Complementing the von Neumann Architecture

Inoue

Pham

2017

SICE Journal of Control, Measurement, and System Integration

Self Cite

View full text Add to dashboard Cite

: Central processing unit (CPU) and graphics processing unit (GPU) are weak ("weak" means inefficiency) at detecting information represented by search, reference and recognition for reason of computer architecture. The memorism processor is a memory base processor which complements CPU's weak point. The two memorism processors called set operating processors (SOP) and database processors (DBP) are the device technology that covers the processing that CPUs and GPUs are not good at. Their profitability in various information processing including rapidity and energy saving performance has been proved and therefore there are great expectations for them as a device technology in the postMoore era. In addition to the conventional SOPs and DBPs, we have developed cross operating processors (XOP), which are excellent at combination/comparison operation and therefore we have applied for a patent of it. These three memorism processors are expected to play a great role in the evolution of the artificial intelligence. This paper is contributed as continuation of [K. Inoue, M. Odaka and C.-K. Pham: Memorism Processor which complements weak point of von Neumann processor, Proc. SII2016, pp. 267-270, 2016, and the authors propose computation that is more suitable for the artificial intelligence era.Key Words : CPU, function-memory, non-von Neumann computer, IoT, AI. Present Situations and Problems of AIIn recent years expectations and interests in artificial intelligence (AI) have been rising as never before. This is because big data have come to be used via internet easily, computing power such as central processing unit (CPU) or graphics processing unit (GPU) has been improved, and the capacity of memories and storage has been expanded.The applicable field of AI is extremely extensive and a number of methods based on it exist in addition. The application fields of AI are extremely wide and there are numerous methods, however the two major trends of AI are deep learning which evolved neural networks and natural language processing [1]. Deep learning imitates the function of the brain by arithmetic operation, especially the high-speed arithmetic operation power such as GPU is the driving force of development. The idea of a neural network as deep learning originates the "arithmetization of thinking" by many mathematicians.As a matter of course, the human brain does not perform arithmetic calculation when the person thinks. Scientists of AI in those days (1950's) only replaced the human thought with the arithmetic calculation which was the maximum value of the computers for convenience. This idea deserves to be called calculism. Although many achievements such as recognition of images and sounds are provided by deep learning, many needs of users for current deep learning are to make learning methods easy, shorten learning time, and further reduce the size and power consumption of the system. If there is another method to * Advanced Original Technologies, 4-7-4-101 Matsubacho, Kashiwashi, Chiba 277-0827, Japan On the oth...

show abstract

Section: Present Situations and Problems Of Aimentioning

confidence: 99%

“…Details of the technique of set operating processors (SOP) based on the above-mentioned definition of pattern are described in reference [3]- [12]. Figure 4 shows a structure of the SOPs which are optimized for processing of stream data including recognition of graphics and character.…”

Section: Outline Of Sopmentioning

confidence: 99%

The Memorism Processor: Towards a Memory-Based Artificially Intelligence Complementing the von Neumann Architecture

Inoue

Pham

2017

SICE Journal of Control, Measurement, and System Integration

Self Cite

View full text Add to dashboard Cite

show abstract

“…Due to this distinct feature, CAM is employed in a wide variety of high-speed searching applications such as image processing and information retrieval. Although there were many approaches to CAM research on FPGAs, dual-port-memory-based (DPM-based) CAM, in our best knowledge, is one of the fastest searching engines so far [12]. The architecture of a 32-word 8-bit DPM-based CAM unit is shown in Fig.…”

Section: Content Addressable Memorymentioning

confidence: 99%

“…For this reason, in this paper, we propose an FPGA-based architecture for fast bitmap indexing. Our design includes a content addressable memory (CAM) whose advantages were fully stated in previous research [12], together with a bit-level transpose matrix. Moreover, the output indexes allow the processing units behind it to answer users' query in parallel using basic logic circuits.…”

Section: Introductionmentioning

confidence: 99%

An FPGA approach for fast bitmap indexing

Nguyen

Pham

2016

IEICE Electron. Express

Self Cite

View full text Add to dashboard Cite

In this paper, an efficient architecture of an FPGA-based bitmap index creation is proposed. The design utilizes a content addressable memory together with a bit-level transpose matrix to index multi-record documents by several given keywords. The experiments in a Cyclone V SX FPGA proved that our circuit could attain throughput of 330.6 million records per second while only using around 71% of embedded memory together with 45% of lookup tables and registers. In fact, achieved throughput is 2.8 times and 1.7 times as high as that of CPU-based and GPU-based design, respectively.

show abstract

“…The Watermarking extraction system takes the advantages of fast and parallel multi-match mode of Content Addressable Memory (CAM). The CAM design based on 2-port memory blocks was presented in our previous work [6]. Fig.…”

Section: A System Designmentioning

confidence: 99%

A fast CAM-based Watermarking extraction on FPGA

Cao²,

Inoue

et al. 2013

Proceedings of 2013 International Conference on IC Design &Amp; Technology (ICICDT)

Self Cite

View full text Add to dashboard Cite

A CAM-based (Content Addressable Memory) system for fast Watermarking extraction has been proposed and implemented on hardware system using FPGA device. The system has a simple structure, does not employ any Central Processor Units (CPUs) or complicated algorithms for extracting hidden information. The authors take advantages of CAM which has an ability of parallel multi-match mode for designing the system. This increases the decryption performance of the Watermarking extraction system. The fast proposed system is applied to fast Watermarking extraction on 2-dimension (2-D) data.

show abstract

A CAM-Based Information Detection Hardware System for Fast Image Matching on FPGA

Cited by 12 publications

References 14 publications

The Memorism Processor: Towards a Memory-Based Artificially Intelligence Complementing the von Neumann Architecture

The Memorism Processor: Towards a Memory-Based Artificially Intelligence Complementing the von Neumann Architecture

An FPGA approach for fast bitmap indexing

A fast CAM-based Watermarking extraction on FPGA

Contact Info

Product

Resources

About