Binary Weighting Research Articles

HARDWARE IMPLEMENTATION OF AN ANALOG SPIKING NEURON WITH DIGITAL CONTROL OF INPUT SIGNALS WEIGHING

Context. Significant challenges facing hardware developers of artificial intelligence systems force them to look for new nonstandard architectural solutions. One of the promising solutions is the transition from von Neumann’s classic architecture to neuromorphic architecture, which at the hardware level tries to imitate the work of the neural network of the human brain. A neuromorphic processor built as hardware implementation of a spiking neural network consists of a large number of elementary electronic circuits that structurally and functionally correspond to neurons. Thus, the design of hardware implementation of a spiking neuron as the basic building element of a neuromorphic processor is of great scientific interest. Objective. The goal of the work is to design an analog spiking neuron hardware implementation with digital control of input signals by binary synaptic weighting coefficients. Method. Designing is performed at the logical/schematic and topological levels of the design flow using modern tools of electronic design automation. All proposed schematic and layout solutions are verified and simulated using computer aided design tools to prove their functionality. Results. The schematic and layout solutions have been developed and investigated for the hardware implementation of the spiking analog neuron with digital control of input signals by binary synaptic weighting coefficients to be the basic building element of a spiking neural network of the neuromorphic processor. Conclusions. The proposed hybrid design of the spiking neuron hardware implementation benefits by combining the simplicity of analog signal processing methods in the neuron with digital control of the state of the neuron using binary weighting coefficients. The simulation results confirm the functionality of the obtained schematic/layout solutions and demonstrate the possibility of implementing logical functions inherent in the perceptron. The prospects for further research may include the design of hardware implementation for a spiking neural network core based on the developed schematic and layout solutions for the spiking neuron.

COALAS II. Extended molecular gas reservoirs are common in a distant, forming galaxy cluster

ABSTRACT This paper presents the results of 475h of interferometric observations with the Australia Telescope Compact Array towards the Spiderweb protocluster at z=2.16. We search for large, extended molecular gas reservoirs among 46 previously detected CO(1−0) emitters, employing a customised method we developed. Based on the CO emission images and position–velocity diagrams, as well as the ranking of sources using a binary weighting of six different criteria, we have identified 14 robust and 7 tentative candidates that exhibit large extended molecular gas reservoirs. These extended reservoirs are defined as having sizes greater than 40 kpc or supergalactic scale. This result suggests a high frequency of extended gas reservoirs, comprising at least 30 percent of our CO-selected sample. An environmental study of the candidates is carried out based on Nth nearest neighbour and we find that the large molecular gas reservoirs tend to exist in denser regions. The spatial distribution of our candidates is mainly centred on the core region of the Spiderweb protocluster. The performance and adaptability of our method are discussed. We found 13 (potentially) extended gas reservoirs located in eight galaxy (proto)clusters from the literature. We noticed that large extended molecular gas reservoirs surrounding (normal) star-forming galaxies in protoclusters are rare. This may be attributable to the lack of observations low-J CO transitions and the lack of quantitative analyses of molecular gas morphologies. The large gas reservoirs in the Spiderweb protocluster are potential sources of the intracluster medium seen in low redshift Virgo- or Coma-like galaxy clusters.

Photonic Spiking Neural Networks with Highly Efficient Training Protocols for Ultrafast Neuromorphic Computing Systems

Photonic technologies offer great prospects for novel, ultrafast, energy-efficient, and hardware-friendly neuromorphic (brain-like) computing platforms. Moreover, neuromorphic photonic approaches based on ubiquitous, technology-mature, and low-cost vertical-cavity surface-emitting lasers (VCSELs) (devices found in fiber-optic transmitters, mobile phones, and automotive sensors) are of particular interest. Given that VCSELs have shown the ability to realize neuronal optical spiking responses (at ultrafast GHz rates), their use in spike-based information-processing systems has been proposed. In this study, spiking neural network (SNN) operation, based on a hardware-friendly photonic system of just one VCSEL, is reported alongside a novel binary weight “significance” training scheme that fully capitalizes on the discrete nature of the optical spikes used by the SNN to process input information. The VCSEL-based photonic SNN was tested with a highly complex multivariate classification task (MADELON) before its performance was compared using a traditional least-squares training method and an alternative novel binary weighting scheme. Excellent classification accuracies of >94% were achieved by both training methods, exceeding the benchmark performance of the dataset in a fraction of the processing time. The newly reported training scheme also dramatically reduces the training set size requirements and the number of trained nodes (≤1% of the total network node count). This VCSEL-based photonic SNN, in combination with the reported “significance” weighting scheme, therefore grants ultrafast spike-based optical processing highly reduced training requirements and hardware complexity for potential application in future neuromorphic systems and artificial intelligence applications.

A content-based recommender system with consideration of repeat purchase behavior

With the increasing popularity of online shopping and information explosion, personalized recommender systems for e-commerce become more and more necessary, which helps customers find the desired products efficiently among variety of categories based on their previous behavior such as buying pattern and rating history. However, most recommender systems for e-commerce adopt binary (purchase/non-purchase) or subjective weighting methods to represent the customer preferences, which is hard to predict their profiles precisely since rapid change in tastes. Therefore, this study focuses on the application of transactional data. A personalized recommender system for e-commerce (PROSE) is proposed in order to enhance the quality of recommendations by integrating the architecture of traditional content-based recommender system with a new component called feedback adjuster, which is designed to make customer implicit feedback reflects the reality of preferences as possible through taking into consideration their behavior of repeat purchase. The computational results indicate that the proposed algorithm is able to outperform other algorithms.

Statistical Wide-Sense Curriculum Learning for Neural Network-Based Pre-Distorter in Coherent Optical Transmitters

Optical communications systems’ performance is limited by physical layer nonlinearities inherent in some of their transmitter components. In order to overcome some of these limitations, several approaches utilizing artificial neural networks to digitally pre-distort the transmitted signal have been proposed in the last few years, alongside other more classical models. However, their performance was strongly dependent on the specific training sequence used. This work improves the performance of neural network-based direct learning approach for digital pre-distortion applications. The proposed method, which is based on the curriculum learning approach, aims to construct a training paradigm of gradually increasing complexity based on the statistical properties of the input signal, in order to achieve better convergence of the network. We prove that the proposed method defines a proper curriculum regardless of the training data distribution, considering the proposed wide-sense curriculum conditions and the case of binary weighting functions. A comparative analysis examines a pre-distorter for a coherent optical transmitter modelled by a Long Short-Term Memory neural network. An improvement of 4.8dB in terms of normalized mean-square error is demonstrated for the case of a 3-bits linear quantizer, compared to the conventional training scheme.

Signal-aware direction-of-arrival estimation using attention mechanisms

The direction-of-arrival (DOA) of sound sources is an essential acoustic parameter used, e.g., for multi-channel speech enhancement or source tracking. Complex acoustic scenarios consisting of sources-of-interest, interfering sources, reverberation, and noise make the estimation of the DOAs corresponding to the sources-of-interest a challenging task. Recently proposed attention mechanisms allow DOA estimators to focus on the sources-of-interest and disregard interference and noise, i.e., they are signal-aware. The attention is typically obtained by a deep neural network (DNN) from a short-time Fourier transform (STFT) based representation of a single microphone signal. Subsequently, attention has been applied as binary or ratio weighting to STFT-based microphone signal representations to reduce the impact of frequency bins dominated by noise, interference, or reverberation. The impact of attention on DOA estimators and different training strategies for attention and DOA DNNs are not yet studied in depth. In this paper, we evaluate systems consisting of different DNNs and signal processing-based methods for DOA estimation when attention is applied. Additionally, we propose training strategies for attention-based DOA estimation optimized via a DOA objective, i.e., end-to-end. The evaluation of the proposed and the baseline systems is performed using data generated with simulated and measured room impulse responses of a uniform-linear microphone array under various acoustic conditions, like reverberation times, noise, and source array distances. The data contains a single source-of-interest, noise, and directional interference. The best-performing systems are also evaluated using measured data. Our experiments show that DNNs used for DOA estimation are biased to the spectral source characteristics and the spectral attention distribution used during training (e.g., spectrally flat/sparse). We also show that this bias in the DOA estimator can be avoided if signal-processing methods are used in combination with attention. Overall, DOA estimation using attention in combination with signal-processing methods exhibits a far lower computational complexity than a fully DNN-based system; however, it yields comparable results.

Noise reduction in infrasound signals based on mask coefficient binary weighting – Generalized cross correlation – Non-negative matrix factorization algorithm

Infrasound noise reduction has two difficulties: the extremely limited datasets and the overlap of frequency bands. To overcome these problems, we present an algorithm, named MCWGCC-NMF, for noise reduction in multi-sensor recordings of mixture infrasound signals. The method is based on non-negative matrix factorization (NMF), and it is combined with spatial information estimated by generalized cross correlation (GCC) and mask coefficient binary weighting (MCW). The source dictionary masking by GCC-NMF, which uses spatial information of individual NMF atoms, is performed by frequency weighting and threshold selection to alleviate the frequency overlap. The results show that the proposed method achieves consistently better performance than the GCC-NMF on the infrasound dataset. Our study provides a signal processing method of infrasound noise reduction, which has the potential to further enhance the signals reception of infrasound arrays with wind noise reduction system (WNRS).

Exploring Feature Definition and Selection for Sentiment Classifiers

In this paper, we systematically explore feature definition and selection strategies for sentiment polarity classification. We begin by exploring basic questions, such as whether to use stemming, term frequency versus binary weighting, negation-enriched features, n-grams or phrases. We then move onto more complex aspects including feature selection using frequency-based vocabulary trimming, part-of-speech and lexicon selection (three types of lexicons), as well as using expected Mutual Information (MI). Using three product and movie review datasets of various sizes, we show, for example, that some techniques are more beneficial for larger datasets than the smaller. A classifier trained on only few features ranked high by MI outperformed one trained on all features in large datasets, yet in small dataset this did not prove to be true. Finally, we perform a space and computation cost analysis to further understand the merits of various feature types.

Connecting the Last.fm Dataset to LyricWiki and MusicBrainz. Lyrics-based experiments in genre classification

Abstract Music information retrieval has lately become an important field of information retrieval, because by profound analysis of music pieces important information can be collected: genre labels, mood prediction, artist identification, just to name a few. The lack of large-scale music datasets containing audio features and metadata has lead to the construction and publication of the Million Song Dataset (MSD) and its satellite datasets. Nonetheless, mainly because of licensing limitations, no freely available lyrics datasets have been published for research. In this paper we describe the construction of an English lyrics dataset based on the Last.fm Dataset, connected to LyricWiki’s database and MusicBrainz’s encyclopedia. To avoid copyright issues, only the URLs to the lyrics are stored in the database. In order to demonstrate the eligibility of the compiled dataset, in the second part of the paper we present genre classification experiments with lyrics-based features, including bagof-n-grams, as well as higher-level features such as rhyme-based and statistical text features. We obtained results similar to the experimental outcomes presented in other works, showing that more sophisticated textual features can improve genre classification performance, and indicating the superiority of the binary weighting scheme compared to tf–idf.

A Performance Evaluation of Attribute Extraction Technique for the Analysis of Sentiment

Sentiment Analysis is the process of computation identifying and categories the thought expressed by people in order to determine whether the writer’s attitude towards a particular product is positive or negative. The results of this analysis can be used in computing customer satisfaction metrics, marketing, contextual advertising, suggestion systems based on the user likes and rating, recommendation systems etc. It can be done on reviews, blogs, tweets, forums etc. Term Weighting is the process of assigning weights to the terms in the documents. Different term weighting schemes like Term Frequency, Term Frequency – Inverse Document Frequency and Binary Weighting methods are used on the reviews to weigh the sentiments words present in the reviews. It is done considering the terms as unigrams, bigrams, trigrams and Syntactic n-grams. The huge data presented to extract sentiments are unstructured in nature and they require processing like classification or clustering to get some meaningful information out of it. Supervised machine learning approaches like Naive Bayes are used to classify the human sentiments present in the reviews as positive and negative.

Establishment of the Performance Evaluation Model of Integrated Water Resources Management and Empirical Study

Based on the socioeconomic development goals of water resources management in a typical river basin of China, this paper puts forward a performance evaluation index model of integrated water resources management (IWRM), classifies the model into four evaluation dimensions by the hierarchical structure of AHP, such as environment sustainability, social fairness, economy benefit and organization energy efficiency, and further divides the four evaluation dimensions into 18 evaluation indices. Besides, this paper establishes a theoretical evaluation system of IWRM performance in different basins on the basis of binary comparison, entropy weighting and linear superposition, and effectively reveals the change trend of IWRM performance in different basins. The results show that the key to IWRM lies in sustainable development of water resources environment, the utilization ratio of water resources and the corresponding economic benefits, and that the rational utilization of water resources is conducive to rapid economic development. From 2003 to 2014, the IWRM performance index in the five basins increased gradually, and the growth rate of the performance indices increased first and then decreased. The change trend of IWRM performance in the five basins is basically the same with that of the performance index.

Reduction of nonlinear embedded boundary models for problems with evolving interfaces

Embedded boundary methods alleviate many computational challenges, including those associated with meshing complex geometries and solving problems with evolving domains and interfaces. Developing model reduction methods for computational frameworks based on such methods seems however to be challenging. Indeed, most popular model reduction techniques are projection-based, and rely on basis functions obtained from the compression of simulation snapshots. In a traditional interface-fitted computational framework, the computation of such basis functions is straightforward, primarily because the computational domain does not contain in this case a fictitious region. This is not the case however for an embedded computational framework because the computational domain typically contains in this case both real and ghost regions whose definitions complicate the collection and compression of simulation snapshots. The problem is exacerbated when the interface separating both regions evolves in time. This paper addresses this issue by formulating the snapshot compression problem as a weighted low-rank approximation problem where the binary weighting identifies the evolving component of the individual simulation snapshots. The proposed approach is application independent and therefore comprehensive. It is successfully demonstrated for the model reduction of several two-dimensional, vortex-dominated, fluid–structure interaction problems.

Compact RGBD Surface Models Based on Sparse Coding

In this paper, we describe a novel approach to construct compact colored 3D environment models representing local surface attributes via sparse coding. Our method decomposes a set of colored point clouds into local surface patches and encodes them based on an overcomplete dictionary. Instead of storing the entire point cloud, we store a dictionary, surface patch positions, and a sparse code description of the depth and RGB attributes for every patch. The dictionary is learned in an unsupervised way from surface patches sampled from indoor maps. We show that better dictionaries can be learned by extending the K-SVD method with a binary weighting scheme that ignores undefined surface cells. Through experimental evaluation on real world laser and RGBD datasets we demonstrate that our method produces compact and accurate models. Furthermore, we clearly outperform an existing state of the art method in terms of compactness, accuracy, and computation time. Additionally, we demonstrate that our sparse code descriptions can be utilized for other important tasks such as object detection.

Low Cost Quasi Binary Weighting Addition Log-SPA LDPC Decoders

This paper proposes a new addition method, which can be applied to reduce the hardware cost of LDPC decoders using log-SPA, which traditionally has the best lowest bit error rate (BER) but the highest hardware requirement. The proposed quasi-binary weighting addition can be simply implemented by OR gates. With auxiliary pseudo-carry circuit, the BER performance can reach a fair level. In the log-SPA message-passing path, numeric transform reduction architecture is proposed for further hardware reduction. Synthesized and numerical results show that the new proposed architecture achieved an up to 32% and 18% total hardware reduction, compared with traditional log-SPA decoders, and the simplest sign-min architecture, respectively, with fair BER performance. DOI: http://dx.doi.org/10.11591/telkomnika.v11i5.2505 Full Text: PDF

Two Types of Distributed CFAR Detection Based on Weighting Functions in Fusion Center for Weibull Clutter

Two types of distributed constant false alarm rate (CFAR) detection using binary and fuzzy weighting functions in fusion center are developed. In the two types of distributed detectors, it was assumed that the clutter parameters at the local sensors are unknown and each local detector performs CFAR processing based on ML and OS CFAR processors before transmitting data to the fusion center. At the fusion center, received data is weighted either by a binary or a fuzzy weighting functions and combined according to deterministic rules, constructing global test statistics. Moreover, for the Weibull clutter, the expression of the weighting functions, based on ML and OS CFAR processors in local detectors, is obtained. In the binary type, we analyzed various distributed detection schemes based on maximum, minimum, and summation rules in fusion center. In the fuzzy type, we consider the various distributed detectors based on algebraic product, algebraic sum, probabilistic OR, and Lukasiewicz t-conorm fuzzy rules in fusion center. The performance of the two types of distributed detectors is analyzed and compared in the homogenous and nonhomogenous situations, multiple targets, or clutter edge. The simulation results indicate the superiority and robust performance of fuzzy type in homogenous and non homogenous situations.

A Selection Model for Soil Retaining Walls in Construction Projects

Recently, there has been an increase in the size, diversification, and specialization of the construction methods of underground sites, particularly due to the need for more efficient utilization of underground spaces in downtown sites where space is limited and land prices high. This study presents a number of procedures that can be used to review site constraints and constructability, and to evaluate the economic efficiency and safety of soil retaining walls in order to select the optimum wall for the given site conditions. Furthermore, this study presents a quantitative analysis model to determine whether the selected soil retaining wall is able to deal with the site risks by using the forced decision method and binary weighting analysis. The study also presents a process model in order to select the optimum soil retaining wall for the given site conditions through the analysis of constructability, economic efficiency, stability and risk factors.

A human motion model based on maps for navigation systems

Foot-mounted indoor positioning systems work remarkably well when using additionally the knowledge of floor-plans in the localization algorithm. Walls and other structures naturally restrict the motion of pedestrians. No pedestrian can walk through walls or jump from one floor to another when considering a building with different floor-levels. By incorporating known floor-plans in sequential Bayesian estimation processes such as particle filters (PFs), long-term error stability can be achieved as long as the map is sufficiently accurate and the environment sufficiently constraints pedestrians' motion. In this article, a new motion model based on maps and floor-plans is introduced that is capable of weighting the possible headings of the pedestrian as a function of the local environment. The motion model is derived from a diffusion algorithm that makes use of the principle of a source effusing gas and is used in the weighting step of a PF implementation. The diffusion algorithm is capable of including floor-plans as well as maps with areas of different degrees of accessibility. The motion model more effectively represents the probability density function of possible headings that are restricted by maps and floor-plans than a simple binary weighting of particles (i.e., eliminating those that crossed walls and keeping the rest). We will show that the motion model will help for obtaining better performance in critical navigation scenarios where two or more modes may be competing for some of the time (multi-modal scenarios).

Improving document clustering using Okapi BM25 feature weighting

We investigate the effect of feature weighting on document clustering, including a novel investigation of Okapi BM25 feature weighting. Using eight document datasets and 17 well-established clustering algorithms we show that the benefit of tf-idf weighting over tf weighting is heavily dependent on both the dataset being clustered and the algorithm used. In addition, binary weighting is shown to be consistently inferior to both tf-idf weighting and tf weighting. We investigate clustering using both BM25 term saturation in isolation and BM25 term saturation with idf, confirming that both are superior to their non-BM25 counterparts under several common clustering quality measures. Finally, we investigate estimation of the k1 BM25 parameter when clustering. Our results indicate that typical values of k1 from other IR tasks are not appropriate for clustering; k1 needs to be higher.

Analysis of a simplified normalized covariance measure based on binary weighting functions for predicting the intelligibility of noise-suppressed speech

The normalized covariance measure (NCM) has been shown previously to predict reliably the intelligibility of noise-suppressed speech containing non-linear distortions. This study analyzes a simplified NCM measure that requires only a small number of bands (not necessarily contiguous) and uses simple binary (1 or 0) weighting functions. The rationale behind the use of a small number of bands is to account for the fact that the spectral information contained in contiguous or nearby bands is correlated and redundant. The modified NCM measure was evaluated with speech intelligibility scores obtained by normal-hearing listeners in 72 noisy conditions involving noise-suppressed speech corrupted by four different types of maskers (car, babble, train, and street interferences). High correlation (r = 0.8) was obtained with the modified NCM measure even when only one band was used. Further analysis revealed a masker-specific pattern of correlations when only one band was used, and bands with low correlation signified the corresponding envelopes that have been severely distorted by the noise-suppression algorithm and/or the masker. Correlation improved to r = 0.84 when only two disjoint bands (centered at 325 and 1874 Hz) were used. Even further improvements in correlation (r = 0.85) were obtained when three or four lower-frequency (<700 Hz) bands were selected.

PARTICLE FILTER WITH BINARY GAUSSIAN WEIGHTING AND SUPPORT VECTOR MACHINE FOR HUMAN POSE INTERPRETATION

Human pose interpretation using Particle filter with Binary Gaussian Weighting and Support Vector Machine isproposed. In the proposed system, Particle filter is used to track human object, then this human object is skeletonizedusing thinning algorithm and classified using Support Vector Machine. The classification is to identify human pose,whether a normal or abnormal behavior. Here Particle filter is modified through weight calculation using Gaussiandistribution to reduce the computational time. The modified particle filter consists of four main phases. First, particlesare generated to predict target’s location. Second, weight of certain particles is calculated and these particles are used tobuild Gaussian distribution. Third, weight of all particles is calculated based on Gaussian distribution. Fourth, updateparticles based on each weight. The modified particle filter could reduce computational time of object tracking sincethis method does not have to calculate particle’s weight one by one. To calculate weight, the proposed method buildsGaussian distribution and calculates particle’s weight using this distribution. Through experiment using video datataken in front of cashier of convenient store, the proposed method reduced computational time in tracking process until68.34% in average compare to the conventional one, meanwhile the accuracy of tracking with this new method iscomparable with particle filter method i.e. 90.3%. Combination particle filter with binary Gaussian weighting andsupport vector machine is promising for advanced early crime scene investigation.