Filter Bank Networks for Few-Shot Class-Incremental Learning

Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the dynamically changing world,e.g.,classifying newly discovered fish species,remains an open problem.We address an even more challenging and realistic setting of this problem where new class samples are insufficient,i.e.,Few-Shot Class-Incremental Learning(FSCIL).Current FSCIL methods augment the training data to alleviate the overfitting of novel classes.By contrast,we propose Filter Bank Networks(FBNs)that augment the learnable filters to capture fine-detailed features for adapting to future new classes.In the forward pass,FBNs augment each convolutional filter to a virtual filter bank containing the canonical one,i.e.,itself,and multiple transformed versions.During back-propagation,FBNs explicitly stimulate fine-detailed features to emerge and collectively align all gradients of each filter bank to learn the canonical one.FBNs capture pattern variants that do not yet exist in the pretraining session,thus making it easy to incorporate new classes in the incremental learning phase.Moreover,FBNs introduce model-level prior knowledge to efficiently utilize the limited few-shot data.Extensive experiments on MNIST,CIFAR100,CUB200,andMini-ImageNet datasets show that FBNs consistently outperformthe baseline by a significantmargin,reporting new state-of-the-art FSCIL results.In addition,we contribute a challenging FSCIL benchmark,Fishshot1K,which contains 8261 underwater images covering 1000 ocean fish species.The code is included in the supplementary materials.

Efficient channelized technology based on filter banks

With the appearance of novel radar signal with low intercept probability, the bandwidth of radar receiver is wider and wider. Wideband digital receiver becomes a research hotspot in the field of communication, radar and electronic reconnaissance, etc. As one of wideband digital receiver systems, digital channelized receiver has become a research emphasis due to the characteristics of full probability receiving and processing multiple signals. Digital channelized technology and signal sampling theory are deeply studied and an efficient channelized model is derived based on filter banks. The correctness of the model is verified by computer simulation The model has less computation compared with the traditional model, which is suitable for engineering application

Efficient design method of non-uniform cosine modulated filter bank for digital hearing aids

To promote the performance of the traditional multichannel filter bank which leads to speech quality degradation,an efficient design method of the non-uniform cosine modulated filter bank（CMFB） based on the audiogram for digital hearing aids is proposed. First, a low-pass prototype filter is designed by the linear iterative algorithm. Secondly,the uniform CMFB is achieved on the basis of the principle formulas. Then, the adjacent channels of a uniform filter bank which have low or gradual slopes are merged according to the trend of audiogram of the hearing impaired person. Finally,the corresponding non-uniform CMFB is obtained. Simulation results show that the signal processed by the proposed filter bank is similar to the original signal in a time-domain waveform and spectrogram without significant distortion or difference. The speech quality results show that the personal evaluation of speech quality（PESQ） of non-uniform CMFB is 35% higher than that of the traditional design, and the hearing-aid speech quality index（HASQI） increases by about 40%.

Designing simultaneous multichannel receivers based on fast filter bank

A scheme to design a simultaneous multichannel receiver is proposed to process multichannel signals in parallel, which is achieved by exploiting the attractive characteristics of a fast filter bank（ FFB）, such as cascaded structure, high frequency selectivity and lowcomputational complexity. Based on the minimization of the objective function, quantified in terms of the total number of multiplications required, subject to prescribed allowable ripples in the passband and stopband, the impulse response coefficients of the prototype filter in each stage are obtained to meet the requirements of the overall specifications for each channel at the receiver side. Simulations and experimental results on the frequency modulation（ FM） broadcast mutlichannel signal receiving system with the FM range from88 to 108 MHz, built upon the proposed FFB structure, are performed to verify its performance. Those results indicate that the proposed scheme is efficient in FM audio indexing applications and has a lower computational complexity, which is approximately 66. 4% of the weighted overlap and add（ WOLA） filter banks based solution.

Adaptive bands filter bank optimized by genetic algorithm for robust speech recognition system

Perceptual auditory filter banks such as Bark-scale filter bank are widely used as front-end processing in speech recognition systems.However,the problem of the design of optimized filter banks that provide higher accuracy in recognition tasks is still open.Owing to spectral analysis in feature extraction,an adaptive bands filter bank (ABFB) is presented.The design adopts flexible bandwidths and center frequencies for the frequency responses of the filters and utilizes genetic algorithm (GA) to optimize the design parameters.The optimization process is realized by combining the front-end filter bank with the back-end recognition network in the performance evaluation loop.The deployment of ABFB together with zero-crossing peak amplitude (ZCPA) feature as a front process for radial basis function (RBF) system shows significant improvement in robustness compared with the Bark-scale filter bank.In ABFB,several sub-bands are still more concentrated toward lower frequency but their exact locations are determined by the performance rather than the perceptual criteria.For the ease of optimization,only symmetrical bands are considered here,which still provide satisfactory results.

Application of uniform DFT filter bank in radar jamming system

The principle of uniform DFT filter bank is presented. Exploiting poly-phase structure, radar jamming system samples the intercepted wideband radar signals through analysis filter bank by different channels and linearly modulates the intercepted radar signal according to the theory of signal and system, then synthesizes the jamming signal through the synthesis filter hank. The method merely requires lower sample frequency, reduces the computational complexity and the data quantity to be processed. The un-ideal filter＇s influence to the result of signals processing is analyzed by simulating the match filter in radar jamming system.

Hardware Architecture of Polyphase Filter Banks Performing Embedded Resampling for Software-Defined Radio Front-Ends

In this paper, we describe resourceefficient hardware architectures for softwaredefined radio （SDR） frontends. These architectures are made efficient by using a polyphase channelizer that performs arbitrary sample rate changes, frequency selection, and bandwidth control. We discuss area, time, and power optimization for field programmable gate array （FPGA） based architectures in an Mpath polyphase filter bank with modified Npath polyphase filter. Such systems allow resampling by arbitrary ratios while simultaneously performing baseband aliasing from center frequencies at Nyquist zones that are not multiples of the output sample rate. A nonmaximally decimated polyphase filter bank, where the number of data loads is not equal to the number of M subfilters, processes M subfilters in a time period that is either less than or greater than the Mdataload ＇ s time period. We present a loadprocess architecture （LPA） and a runtime architecture （RA） （based on serial polyphase structure） which have different scheduling. In LPA, Nsubfilters are loaded, and then M subfilters are processed at a clock rate that is a multiple of the input data rate. This is necessary to meet the output time constraint of the down-sampled data. In RA, Msubfilters processes are efficiently scheduled within Ndataload time while simultaneously loading N subfilters. This requires reduced clock rates compared with LPA, and potentially less power is consumed. A polyphase filter bank that uses different resampling factors for maximally decimated, underdecimated, overdecimated, and combined upand downsampled scenarios is used as a case study, and an analysis of area, time, and power for their FPGA architectures is given. For resourceoptimized SDR frontends, RA is superior for reducing operating clock rates and dynamic power consumption. RA is also superior for reducing area resources, except when indices are prestored in LUTs.

Polyphase Filter Banks for Embedded Sample Rate Changes in Digital Radio Front-Ends

This paper presents efficient processing engines for software-defined radio （SDR） front-ends. These engines, based on a polyphase channelizer, perform arbitrary sample-rate changes, frequency selection, and bandwidth control. This paper presents an M-path polyphase filter bank based on a modified N-path polyphase filter. Such a system allows resampling by arbitrary ratios while performing baseband aliasing from center frequencies at Nyquist zones that are not multiples of the output sample rate. This resampling technique is based on sliding cyclic data load interacting with cyclic-shifted coefficients. A non-maximally-decimated polyphase filterbank （where the number of data loads is not equal to the number of M subfilters） processes M subfilters in a time period that is less than or greater than the M data loads. A polyphase filter bank with five different resampling modes is used as a case study for embedded resamp/ing in SDR front-ends. These modes are （i） maximally decimated, （ii） Under-decimated, （iii） over-decimated, and combined up- and down-sampling with （iv） single stride length, and （v） multiple stride lengths. These modes can be used to obtain any required rational sampling rate change in an SDR front-end based on a polyphase channelizer. They can also be used for translation to and from arbitrary center frequencies that are unrelated to the output sample rates.

Digital Channelization Technology for HF Communication Base on Fast Filter Bank

Channelization typically realized by digital filter banks is an important topic in high frequency(HF) communication and software defined radios(SDR) areas. Channelization has a rigorous requirement for the characteristic of frequency response, e.g., steep transitional band and sharp decay. To address this issue, we investigated the feasibility and implementation of applying fast filter bank(FFB) in channelization in this paper. We analyzed the butterfly structure of FFB similar with fast Fourier transform(FFT), in which prototype sub-filters are cascaded to achieve a low complexity. Hence, it is suitable for designing filter bank with steep transitional band and sharp decay in stop-band. Moreover, we designed a pipelined structure of FFB to achieve a balance between area and performance. Design example shows that FFB has lower computational complexity compared with the other filter banks.

Research progress in SAW filter banks

SAW （Surface Acoustic Wave） filter bank is a single input, single or multi-output device consisting of multi-SAW-filters with input interconnection network or switch circuits, and can be divided into two categories： channelizer （multi-output） and switchable （programmable, single output）. The tbrmer is mainly used in military channelized receiver for spectrum analysis; the latter has wide application in frequency synthesizer and frequency-hopping radar and communication system receiver as anti-jamming filter, and has been widely used in various military electronic equipments ever since the 1970s. Research abroad was done mainly by Americans, few documents on related work done by Japan and Russia are available. Domestic research started in the 1980s, mainly by No. 26 Research Institute, China Electronics Technology Group Co., Institute of Acoustics, Chinese Academy of Sciences, No. 23 and No. 25 Research Institute, China Spaceflight Tech. Group Co. This paper first briefly introduces Chinese and foreign research on SAW filter banks; then discusses research progress in device design, the input interconnection network or switch circuit and miniaturization; and ends in a brief perspective of developing trends in future.

Directions of Arrival Optimization Method for Broadband Antenna Array Based on Filter Bank

This paper proposes an estimation method of the directions of arrival(DOA) for the broadband signals received by a uniform planar antenna array including the mutual coupling effect. The process modeling of antenna array receiving dynamic signal is obtained accurately via building the array system electromagnetic space-time model, and the mutual coupling effect is included in the synthesis procedure automatically. The broadband signal is divided into several sub-bands, and the complex weight of these sub-bands frequencies is figured out using the least mean square iteration programming. In the proposed method, the mutual coupling effect is compensated at the different frequencies. The calculation results show that the method is suitable for the uniform linear and uniform circular array pattern integration, and it is commendably satisfied with the requirement of engineering design.

Perfect Reconstructable Decimated One-Dimensional Empirical Mode Decomposition Filter Banks

This paper introduces decimated filter banks for the one-dimensional empirical mode decomposition （1D-EMD）. These filter banks can provide perfect reconstruction and allow for an arbitrary tree structure. Since the EMD is a data driven decomposition, it is a very useful analysis instrument for non-stationary and non-linear signals. However, the traditional 1D-EMD has the disadvantage of expanding the data. Large data sets can be generated as the amount of data to be stored increases with every decomposition level. The 1D-EMD can be thought as having the structure of a single dyadic filter. However, a methodology to incorporate the decomposition into any arbitrary tree structure has not been reported yet in the literature. This paper shows how to extend the 1D-EMD into any arbitrary tree structure while maintaining the perfect reconstruction property. Furthermore, the technique allows for downsampling the decomposed signals. This paper, thus, presents a method to minimize the data-expansion drawback of the 1D-EMD by using decimation and merging the EMD coefficients. The proposed algorithm is applicable for any arbitrary tree structure including a full binary tree structure.

Faster-than-Nyquist signaling based on filter bank multicarrier modulation with joint optimization

Multi-carrier faster-than-Nyquist(MFTN)can improve the spectrum efficiency(SE).In this paper,we first analyze the benefit of time frequency packing MFTN(TFP-MFTN).Then,we propose an efficient digital implementation for TFP-MFTN based on filter bank multicarrier modulation.The time frequency packing ratio pair in our proposed implementation scheme is optimized with the SE criterion.Next,the joint optimization for the coded modulation MFTN based on extrinsic information transfer(EXIT)chart is performed.The Monte-Carlo simulations are carried out to verify performance gain of the joint inner and outer code optimization.Simulation results demonstrate that the TFPMFTN has a 0.8 dB and 0.9 dB gain comparing to time packing MFTN(TP-MFTN)and higher order Nyquist at same SE,respectively;the TFP-MFTN with optimized low density parity check(LDPC)code has a 2.9 dB gain comparing to that with digital video broadcasting(DVB)LDPC.Compared with previous work on TFP-MFTN(SE=1.55 bit/s/Hz),the SE of our work is improved by 29%and our work has a 4.1 dB gain at BER=1×10^(-5).

Vessel Extraction on Ocular Fundus Images by Using Gabor Filter Bank

Fundus diagnosis is an important part of the whole body examination that may provide rich clinical information to doctors for diagnostic reference. Manual fundus vessel extraction is helpful to quantitative measurement of diseases but obviously it is a tough work for physicians. This paper presents an automatic method by using Gabor filter bank to extract the artery and vein separately in the ocular fundus images. After preprocessing steps that include gray-scale transform, gray value inversion and contrast enhancement, the Gabor filter bank is applied to the extraction of the artery and vein in the ocular fundus images. Finally these two different width types of vessels are selected by post-processing methods such as labeling, corrosion, binarization, etc. Evaluation results show an accurate rate of 90% in vein and 82% in artery from 20 cases, that indicates the effectiveness of our proposed segmentation method.

A Multiobjective Optimization Method for Designing <i>M</i>-Channel NPR Cosine Modulated Filter Bank for Image Compression

This paper proposes a method to design multichannel cosine modulated filter bank for image compression using multiobjective optimization technique. The design problem is a combination of stopband residual energy, least square error of the overall transfer function of the filter bank, coding gain with dc leakage free condition as constraint. The proposed algorithm uses Non-dominated Sorting Genetic Algorithm (NSGA) to minimize the mutually contradictory objective function by minimizing filter tap weights of prototype filter. The algorithm solves this problem by searching solutions that achieve the best compromise between the different objectives criteria. The performance of this algorithm is evaluated in terms of coding gain and peak signal to noise ratio (PSNR). Simulation results on different images are included to illustrate the effectiveness of the proposed algorithm for image compression application.

Image Encryption Scheme Based on Filter Bank and Lifting

In this paper, the quality of image encryption using filter bank with lifting scheme has been studied and evaluated. Many extensive techniques have been applied to examine the security of the image encryption using the filter bank cipher with one or two rounds. To analyze and evaluate the proposed scheme, many parameters have been employed, such as, histogram analysis, correlation coefficient, global entropy, block entropy, avalanche effect, number of pixel change rate (NPCR), unified average change intensity (UACI), compression friendliness, exhaustive key analysis, and key sensitivity test. The simulation results showed that, the quality of the image encryption passes all these tests. Moreover, it reaches or excels the current state-of-the-arts. So that, the proposed image encryption process produces random-like output from the statistical point of views.

THEORY AND DESIGN OF UNIFORM FILTER BANKS USING ALL-PASS FILTERS

In this paper,the theory of uniform filter banks using all-pass tilters is furtherdeveloped.A new structure of two stage filter banks using all-pass filter is proposed,The pre-stage is half-band filter with period,the post-stage is two sets of band-pass filter banks.Thepre-stage filter stop-band just controls the don’t-care-band of the post-stage filter banks usingall-pass polyphase,so as to realize a continuous stop-band property Moreover,a method ofsynthesizing filter bank is derived,which eliminates aliasing and amplitude distortions of theanalysis/synthesis system Finally,an example is given.

A NEW DESIGN METHOD OF CDMA SPREADING CODES BASED ON MULTI-RATE UNITARY FILTER BANK

It is well-known that the multi-valued CDMA spreading codes can be designed by means of a pair of mirror multi-rate filter banks based on some optimizing criterion. This paper indicates that there exists a theoretical bound in the performance of its circulating correlation property, which is given by an explicit expression. Based on this analysis, a criterion of maximizing entropy is proposed to design such codes. Computer simulation result suggests that the resulted codes outperform the conventional binary balanced Gold codes for an asynchronous CDMA system.

A Non-Maximally Decimated Dynamic Reconfigurable Channelized Structure Based on Modulated Filter Bank

A structure of dynamic reconfigurable channelized filter bank is proposed in order to solve the problem that the uniform channelized receiver cannot receive the cross-channel and wideband signal. The dynamic reconfigurable channelized filter bank is divided into two parts-the analysis filter bank and the synthesis filter bank. The function of the analysis filter bank is to divide the received signal into several sub-signals according to the channel division. Then the sub-signals of each channel need to be detected and discriminated. At last, we use the sub-signals to reconstruct the original received signal by the synthesis filter bank. The analysis filter and the synthesis filter bank of the dynamic reconfigurable channelized filter bank are all efficient polyphase structures, so it can save more hardware resources and has extensive applicability. The structure is simulated by MATLAB and the simulation results verify the correctness of this structure.

Design of Dynamic Reconfigurable Structure Based on Integrated Filter Banks

In electronic confrontation, radar confrontation is an important part. Various radars widely used in modern warfare are the most important equipment in the field of information acquisition and precision guidance. Especially in the vast battle space, in order to achieve timely, accurate and comprehensive access to various target information, the role of radar is irreplaceable. Especially in the vast battle space, the role of radar is irreplaceable in order to achieve timely, accurate and comprehensive access to various target information. Therefore, in the war, it will not be able to protect their own survival and play combat effectiveness if not having the radar's ability to fight. At this time, that is the need for radar equipment in time to detect radar exposure, rapid measurement of radar signal parameters and identify threat signals, and targeted to do the implementation of interference or the implementation of technical attacks. In order to extract the parameters of the received signal to facilitate the subsequent research and analysis, this paper deduces the structure of the integrated filter bank in detail, and gives the reconfigurable filter bank structure design method, under the condition of accurate reconstruction of the signal. Based on the analysis of the design and calculation complexity of the filter bank structure, the dynamic reconfigurable design method con-sumes less hardware resources and wide application range, and the simulation structure also verifies the correctness and flexibility of the structure.