Variable Word Length Research Articles

FlexDML: High Utilization Configurable Multimode Arithmetic Units Featuring Dual Mode Logic

Dual Mode Logic (DML) enables flexible energy-delay optimization. By setting the design elements in static or dynamic mode, it is possible to significantly improve performance at the cost of a limited increase in energy consumption, and vice versa. In this work, for the first time, we design, fabricate in a commercial 16nm FinFET process, and evaluate in silicon FlexDML, an adaptively configurable arithmetic unit. FlexDML comprises several DMLBricks, 2×2 and 4×4 DML-based multiplier units, as well as DML-based shifters and adders. FlexDML is capable of performing several arithmetic operations with flexible wordlength, such as N×N-bit, two (N/2)×N-bit or N×(N/2)-bit, four (N/2)×(N/2)-bit and so on multiplications, as well as shifts and inner product calculations. All DMLBricks can be operational at the same time, increasing the hardware utilization to the level unattainable in typical designs that support variable wordlength. FlexDML is the first to support a pipeline multimode, where the DML mode is set separately for each pipeline stage, thus enabling more flexible energy-delay optimization. FlexDML achieves up to 41% performance improvement in dynamic mode, and power savings of about 23% in static mode, compared to CMOS implementation.

Statistical word segmentation: Anchoring learning across contexts.

The present experiments were designed to assess infants' abilities to use syllable co-occurrence regularities to segment fluent speech across contexts. Specifically, we investigated whether 9-month-old infants could use statistical regularities in one speech context to support speech segmentation in a second context. Contexts were defined by different word sets representing contextual differences that might occur across conversations or utterances. This mimics the integration of information across multiple interactions within a single language, which is critical for language acquisition. In particular, we performed two experiments to assess whether a statistically segmented word could be used to anchor segmentation in a second, more challenging context, namely speech with variable word lengths. The results of Experiment 1 were consistent with past work suggesting that statistical learning may be hindered by speech with word-length variability, which is inherent to infants' natural speech environments. In Experiment 2, we found that infants could use a previously statistically segmented word to support word segmentation in a novel, challenging context. We also present findings suggesting that this ability was associated with infants' early word knowledge but not their performance on a cognitive development assessment.

Performance Improvement of Vector-Radix Decimation-in-Frequency 3D-DCT/IDCT Using Variable Word Length

High-speed video coding and compression are extensively used in many IoT applications with optimum data usage and resolution using three-dimensional discrete cosine transforms (3D-DCT). We propose an efficient hardware implementation for high-speed vector-radix decimation-in-frequency (VR-DIF) 3D-DCT with an optimum area and power consumption. In the previous implementation, the data path arithmetic units used a fixed word length (either 16 or 18 or 21 bits), whereas the proposed architecture uses the range of word length from 11 bits (1-bit sign, 1-bit integer and 9-bit fraction) to 20 bits (1-bit sign, 10-bit integer and 9-bit fraction) to achieve lower silicon area and power consumption. The architecture is optimally pipelined to achieve high processing speed (above 3 Giga samples/s). To test the proposed architecture, an $$8\times 8\times 8$$ video cube with a pixel depth of 8 bits is considered. The arithmetic functional units such as signed adder/subtractor and cosine coefficient multipliers required for implementing $$8\times 8\times 8$$ 3D-DCT/IDCT processor is designed with the proposed variable word length. The core of VR-DIF 3D-DCT/IDCT with the variable word length is implemented using TSMC 90 nm technology library. The proposed architecture consumes 26.5% and 23.2% lesser area and power, respectively, than the existing fixed word length 3D-DCT-II implementation tested with a maximum frequency of 653 MHz.

Discriminative Dictionary Learning for Time Series Classification

Time series symbolization based on the Symbolic Fourier Approximation (SFA) and a sliding window mechanism can effectively improve classification performance. Hence, it has become a research hotspot of time series representation learning. However, there are still some obvious shortcomings. First, the lengths of the generated words are consistent for different sliding windows in the symbolization process, ignoring the difference of the discriminative information contained in diverse periods. Second, research on the relationship between words is limited to adjacent ones. Third, existing dictionary learning methods do not consider a filtering algorithm for discriminative word selection. To this end, a novel and fast variable-length word generation method, incorporating the skip-bigram model for the construction of symbiotic word pairs, is proposed for the first time in our work. Then, a discriminative word filter with a dynamic threshold is designed to build the discriminative word dictionary. Many controlled experiments first verified the effectiveness of each proposed method. Then, the performance improvement ability of the word dictionary is demonstrated by the comparative experiments with the representative classification models based on different theoretical foundations.

Compact and high‐throughput parameterisable architectures for memory‐based FFT algorithms

Designers must carefully choose the best-suited fast Fourier transform (FFT) algorithm among various available techniques for the custom implementation that meets their design requirements, such as throughput, latency, and area. This article, to the best of authors' knowledge, is the first to present a compact and yet high-throughput parameterisable hardware architecture for implementing different FFT algorithms, including radix-2, radix-4, radix-8, mixed-radix, and split-radix algorithms. The designed architectures are fully parameterisable to support a variety of transform lengths and variable word-lengths. The FFT algorithms have been modelled and simulated in double-precision floating-point and fixed-point representations using authors' custom-developed library of numerical operations. The designed FFT architectures are modelled in Verilog hardware description language and their cycle-accurate and bit-true simulation results are verified against their fixed-point simulation models. The characteristics and implementation results of various FFT architectures on a Xilinx Virtex-7 FPGA are presented. Compared to recently published works, authors' memory-based FFT architectures utilise less reconfigurable resources while maintaining comparable or higher operating frequencies. The ASIC implementation results in a standard 45-nm CMOS technology are also presented for the designed memory-based FFT architectures. The execution times of FFTs on a workstation and a graphics processing unit are compared against authors' FPGA implementations.

Properties and Design of Variable-to-Variable Length Codes

For the entropy coding of independent and identically distributed (i.i.d.) binary sources, variable-to-variable length (V2V) codes are an interesting alternative to arithmetic coding. Such a V2V code translates variable length words of the source into variable length code words by employing two prefix-free codes. In this article, several properties of V2V codes are studied, and new concepts are developed. In particular, it is shown that the redundancy of a V2V code cannot be zero for a binary i.i.d. source {X} with 0 < p X (1) < 0.5. Furthermore, the concept of prime and composite V2V codes is proposed, and it is shown why composite V2V codes can be disregarded in the search for particular classes of minimum redundancy codes. Moreover, a canonical representation for V2V codes is proposed, which identifies V2V codes that have the same average code length function. It is shown how these concepts can be employed to greatly reduce the complexity of a search for minimum redundancy (size-limited) V2V codes.

Um polvo é igual a um pato? Sobre o impacto das variáveis fonológicas na avaliação do conhecimento metalinguístico

Most tools used to assess phonological awareness (PA) are not controlled from a phonological perspective. However, research available for different languages refers the impact of phonological variables in the assessment of this type of linguistic knowledge. In this paper, we will be focused on the impact of the prosodic variables word length, syllable constituency and word stress in the performance of tasks assessing Portuguese children’s PA. A sample of 49 monolingual and typically developing Portuguese children was gathered (22 first graders and 27 third graders). Three tasks were used to assess syllable awareness (1 segmentation task, 1 identification task, 1 omission task); one segmentation task assessed the children’s segmental awareness; another segmentation task assessed their awareness of syllable constituency. Although some studies report the effect of all prosodic variables mentioned above (see state of the art below), the results in this paper show a significant effect of syllable structure, lexical stimuli with branching onsets and branching rhymes being the most problematic, but no significant effects of word length or word stress. The results are an empirical contribution for the discussion of the aspects that need to be taken in to account to build PA assessment tools and to plan intervention in clinical and in educational contexts

Dynamic Reconfigurable Ternary Content Addressable Memory for OpenFlow-Compliant Low-Power Packet Processing

OpenFlow, the main protocol for software-defined networking (SDN) advocated by Google, requires multiple flow tables with long and variable rule lengths. For fast packet forwarding, using ternary content-addressable memory (TCAM) as implementation of lookup-table has displayed superior performance in executing the conventional packet classification algorithms. However, applying traditional TCAM designs to an OpenFlow switch leads to problems of insufficient word length, inflexible table configuration, and high-power consumptions. This paper presents a low-power dynamic reconfigurable TCAM design for OpenFlow-compliant packet processing. It can support long and variable word lengths for different flow tables, with less than 5% of area overhead. To reduce dynamic power consumption without sacrificing throughput, the proposed hybrid-pipelined structure based on NAND-NOR CAM and pipelined-TCAM, can achieve lower energy-delay products than traditional TCAM. Moreover, the proposed self-power gating technique controlled by the mask data optimizes both NAND and NOR core cells, which lowers power consumption without additional control and routing overhead. Finally, the low-power reconfigurable TCAM is implemented in TSMC 40 nm CMOS technology. The implementation results show that the energy metric of the proposed reconfigurable TCAM is only 0.177 fJ/bit/search, at a 416 MHz operating frequency. Thus, the proposed TCAM design is very suitable for the emerging high performance packet processing in future SDN applications that need long and variable word lengths.

On the comparison of regulatory sequences with multiple resolution Entropic Profiles.

BackgroundEnhancers are stretches of DNA (100–1000 bp) that play a major role in development gene expression, evolution and disease. It has been recently shown that in high-level eukaryotes enhancers rarely work alone, instead they collaborate by forming clusters of cis-regulatory modules (CRMs). Although the binding of transcription factors is sequence-specific, the identification of functionally similar enhancers is very difficult and it cannot be carried out with traditional alignment-based techniques.ResultsThe use of fast similarity measures, like alignment-free measures, to detect related regulatory sequences is crucial to understand functional correlation between two enhancers. In this paper we study the use of alignment-free measures for the classification of CRMs. However, alignment-free measures are generally tied to a fixed resolution k. Here we propose an alignment-free statistic, called EP^{*}_{2}, that is based on multiple resolution patterns derived from the Entropic Profiles (EPs). The Entropic Profile is a function of the genomic location that captures the importance of that region with respect to the whole genome. As a byproduct we provide a formula to compute the exact variance of variable length word counts, a result that can be of general interest also in other applications.ConclusionsWe evaluate several alignment-free statistics on simulated data and real mouse ChIP-seq sequences. The new statistic, EP^{*}_{2}, is highly successful in discriminating functionally related enhancers and, in almost all experiments, it outperforms fixed-resolution methods. We implemented the new alignment-free measures, as well as traditional ones, in a software called EP-sim that is freely available: http://www.dei.unipd.it/~ciompin/main/EP-sim.html.

Power Reduction of Variable Wordlength OFDM Receiver in Time-Varying Fading Channels by Monitoring Subcarrier SNRs

Determination of wordlength is essential for designing digital circuits because the wordlength affects system performance, hardware size, and power consumption. Variable wordlength methods that a system dynamically and effectively changes the wordlength depending on surrounding environments have been studied for power reduction in wireless systems. The conventional variable wordlength methods induce communication performance degradation when compared with a floating-point representation in time-varying fading channels. This paper discusses rapid wordlength control on packet basis and proposes a new method based on monitoring subcarrier SNRs in an OFDM receiver. The proposed method can estimate signal quality accurately and can decrease the wordlength decision errors. The simulation results have indicated that the proposed method shows better PER performance compared with the conventional methods.

A Probabilistic Approach to Generalized Zeckendorf Decompositions

Generalized Zeckendorf decompositions are expansions of integers as sums of elements of solutions to recurrence relations. The simplest cases are base-$b$ expansions, and the standard Zeckendorf decomposition uses the Fibonacci sequence. The expansions are finite sequences of nonnegative integer coefficients (satisfying certain technical conditions to guarantee uniqueness of the decomposition) and which can be viewed as analogues of sequences of variable-length words made from some fixed alphabet. In this paper we present a new approach and construction for uniform measures on expansions, identifying them as the distribution of a Markov chain conditioned not to hit a set. This gives a unified approach that allows us to easily recover results on the expansions from analogous results for Markov chains, and in this paper we focus on laws of large numbers, central limit theorems for sums of digits, and statements on gaps (zeros) in expansions. We expect the approach to prove useful in other similar contexts.

Sentiment analysis via integrating distributed representations of variable-length word sequence

Sentiment analysis aims to identify the overall emotional polarity of a given text. It is a nontrivial task to perform sentiment analysis as sentiment information is crucial in many natural language processing applications. Previous n-gram features is derived from a bag-of-n-gram model which is insensitive to the order of the n-gram. To address this problem, we integrates distributed semantic features of word sequence, with fixed-size independent of the length of the word sequence. We also learn distributed semantic features of part-of-speech (POS) sequence as additional syntax-related clues to sentiment analysis. Our semantic features are able to capture both local contexts and global contexts automatically without involving comprehensive task-specific feature engineering. We validate the effectiveness of the method on our constructed sentiment dataset. Experiment results show that our method are able to improve the quality of sentiment analysis when comparing with several competitive baselines.

Beyond Fixed-Resolution Alignment-Free Measures for Mammalian Enhancers Sequence Comparison.

The cell-type diversity is to a large degree driven by transcription regulation, i.e., enhancers. It has been recently shown that in high-level eukaryotes enhancers rarely work alone, instead they collaborate by forming clusters of cis-regulatory modules (CRMs). Even if the binding of transcription factors is sequence-specific, the identification of functionally similar enhancers is very difficult. A similarity measure to detect related regulatory sequences is crucial to understand functional correlation between two enhancers. This will allow large-scale analyses, clustering and genome-wide classifications. In this paper we present Under2, a parameter-free alignment-free statistic based on variable-length words. As opposed to traditional alignment-free methods, which are based on fixed-length patterns or, in other words, tied to a fixed resolution, our statistic is built upon variable-length words, and thus multiple resolutions are allowed. This will capture the great variability of lengths of CRMs. We evaluate several alignment-free statistics on simulated data and real ChIP-seq sequences. The new statistic is highly successful in discriminating functionally related enhancers and, in almost all experiments, it outperforms fixed-resolution methods. Finally, experiments on mouse enhancers show that Under2 can separate enhancers active in different tissues. http://www.dei.unipd.it/~ciompin/main/UnderIICRMS.html.

Variable-Length Constrained Sequence Codes

We outline the construction of simple, high rate constrained sequence codes with variable-length source words and variable-length codewords based on the search of partial extension of a minimal set of codeword lengths. We present optimal word length mappings for a subset of constrained sequence codes, and provide examples of (d, k) and DC-free codes with few codewords and average code rates greater than 99% of capacity.

Improved SIMD Architecture for High Performance Video Processors

Single instruction multiple data (SIMD) execution is in no doubt an efficient way to exploit the data level parallelism in image and video applications. However, SIMD execution bottlenecks must be tackled in order to achieve high execution efficiency. We first analyze in this paper the implementation of two major kernel functions of H.264/AVC namely, SATD and subpel interpolation, in conventional SIMD architectures to identify the bottlenecks in traditional approaches. Based on the analysis results, we propose a new SIMD architecture with two novel features: 1) parallel memory structure with variable block size and word length support, and 2) configurable SIMD structure. The proposed parallel memory structure allows great flexibility for programmers to perform data access of different block sizes and different word lengths. The configurable SIMD structure allows almost “random” register file access and slightly different operations in ALUs inside SIMD. The new features greatly benefit the realization of H.264/AVC kernel functions. For instance, the fractional motion estimation, particularly the half to quarter pixel interpolation, can now be executed with minimal or no additional memory access. When comparing with the conventional SIMD systems, the proposed SIMD architecture can have a further speedup of 2.1X to 4.6X when implementing H.264/AVC kernel functions. Based on Amdahl's law, the overall speedup of H.264/AVC encoding application can be projected to be 2.46X. We expect significant improvement can also be achieved when applying the proposed architecture to other image and video processing applications.

Variable wordlength soft-decision Viterbi decoder for power-efficient wireless LAN

This paper describes the design of a new, variable wordlength, soft-decision Viterbi decoder that can significantly reduce power dissipation in wireless local area network (LAN) hardware. By taking into account the dynamic range of the time-varying channel coefficients, the quantization level is adjusted to suit the observed instantaneous link quality, i.e., the decoder wordlength is adjusted to maintain efficient throughput requirements and low power dissipation. A dynamic voltage scaling is applied with a variable wordlength to significantly reduce power consumption in the soft-decision Viterbi decoder. Unlike the conventional 8-bit fixed-wordlength decoder, our radix-4 Viterbi decoder can reduce power consumption by up to 35% under AWGN and multipath fading channel conditions.

Improved accuracy of supervised CRM discovery with interpolated Markov models and cross-species comparison

Despite recent advances in experimental approaches for identifying transcriptional cis-regulatory modules (CRMs, ‘enhancers’), direct empirical discovery of CRMs for all genes in all cell types and environmental conditions is likely to remain an elusive goal. Effective methods for computational CRM discovery are thus a critically needed complement to empirical approaches. However, existing computational methods that search for clusters of putative binding sites are ineffective if the relevant TFs and/or their binding specificities are unknown. Here, we provide a significantly improved method for ‘motif-blind’ CRM discovery that does not depend on knowledge or accurate prediction of TF-binding motifs and is effective when limited knowledge of functional CRMs is available to ‘supervise’ the search. We propose a new statistical method, based on ‘Interpolated Markov Models’, for motif-blind, genome-wide CRM discovery. It captures the statistical profile of variable length words in known CRMs of a regulatory network and finds candidate CRMs that match this profile. The method also uses orthologs of the known CRMs from closely related genomes. We perform in silico evaluation of predicted CRMs by assessing whether their neighboring genes are enriched for the expected expression patterns. This assessment uses a novel statistical test that extends the widely used Hypergeometric test of gene set enrichment to account for variability in intergenic lengths. We find that the new CRM prediction method is superior to existing methods. Finally, we experimentally validate 12 new CRM predictions by examining their regulatory activity in vivo in Drosophila; 10 of the tested CRMs were found to be functional, while 6 of the top 7 predictions showed the expected activity patterns. We make our program available as downloadable source code, and as a plugin for a genome browser installed on our servers.

Learning orthographic and phonological representations in models of monosyllabic and bisyllabic naming

Most current models of word naming are restricted to processing monosyllabic words and pseudowords. This limitation stems from difficulties in representing the orthographic and phonological codes for words varying substantially in length. Sibley, Kello, Plaut, and Elman (2008) described an extension of the simple recurrent network architecture, called the sequence encoder, that learned orthographic and phonological representations of variable-length words. The present research explored the use of sequence encoders in models of monosyllabic and bisyllabic word naming. Performance in these models is comparable to other models in terms of word and pseudoword naming accuracy, as well as accounting for naming latency phenomena. Although the models do not address all naming phenomena, the results suggest that sequence encoders can learn orthographic and phonological representations, making it easier to create models that scale up to larger vocabularies, while accounting for behavioural data.

Processos de substituição e variabilidade articulatória na fala de sujeitos com dispraxia verbal

O objetivo do presente estudo foi analisar o papel das variáveis linguísticas na ocorrência dos processos de substituição na fala de sujeitos com dispraxia verbal (DV). Para isso, foi realizada a análise fonológica de sete sujeitos com idades entre 2:6 (anos:meses) e 4:2, com hipótese diagnóstica de DV. As ocorrências dos processos de substituições usuais e idiossincráticas, de assimilações e de variabilidade articulatória foram analisadas estatisticamente por meio do pacote computacional VARBRUL. A variável extensão da palavra foi estatisticamente significante para a ocorrência de assimilações e substituições não usuais, indicando que as variantes trissilábicas e polissilábicas foram as maiores favorecedoras de ocorrência desses processos. A tonicidade foi estatisticamente significante para a ocorrência da variabilidade articulatória e substituições usuais, sendo que o processo apresentou maior probabilidade de ocorrência em sílaba tônica e pós-tônica (sílabas dentro do pé métrico do acento), respectivamente. A classe de sons foi significativa para a realização de substituições usuais pelos sujeitos estudados, ocorrendo quando os segmentos são fonemas líquidos e fricativos. Por fim, a estrutura silábica foi estatisticamente significante para as substituições idiossincráticas. As posições de coda final e de onset simples medial foram as mais suscetíveis à ocorrência do processo. Os dados desta pesquisa sugerem que as substituições, de uma forma geral, tendem a ocorrer em palavras com mais de duas sílabas, em alvos líquidos e fricativos, dentro do pé-métrico do acento (em tônica e pós-tônica), em posição de onset simples medial e coda final.

Formula omitted]-difference matching in amortized linear time for all the words in a text

Given a text x of length n , we study the problem of solving the k -difference problem for all the words, either with fixed or variable length, taken from the text itself. The result finds its application in pattern discovery in biosequences where over- or under-represented words are extracted from the input sequences. The proposed algorithm runs in amortized linear time per word. This improves the complexity obtained by applying well-known algorithms to each of the O ( n ) fixed length words or O ( n 2 ) variable length words in x by factor of k , k log k , or m log m , depending on the chosen algorithm. The space required is O ( n ) if we just count the occurrences, or O ( n 2 ) if we also store the positions. This second scenario can be used as the basis for other applications, such as searching gapped factors with mismatches or approximate pattern matching extended to any word.