Basic Processing Unit Research Articles

Strategies for unsupervised multimedia processing

In this article, we explore a new family of neural network architectures that have a basis in self-organization, yet are somewhat free from many of the constraints typical of other well-known self-organizing architectures. Within this family, the basic processing unit is known as the self-organizing tree map (SOTM). We will look at how this model has evolved since its inception in 1995, how it has inspired new models, and how it is being applied to complex multimedia research problems in digital asset management and microbiological image analysis.

Quantum Computing & Information Technology: A Tutorial

The quantum computing and information technology has emerged as an important area of research and development. It is a promising field as it is believed to solve many problems here to not possible with classical computer. The objective of this article is to present a review and tutorial of quantum computing, quantum teletransport and quantum security. It gives the basic introduction of the subject, compares aspects with those of classical computers in terms of basic information processing units, gates, key exchange and reports practical designs available in the literature.

Signal processing for in-car communication systems

Due to a large amount of background noise the communication within a car driving at high or even moderate speed is often difficult. This is especially true if one of the communication partners is the driver and the other is one of the backseat passengers. As a result of the high noise level the backseat passengers often lean towards the front passengers. Furthermore, all speakers increase their loudness. Even if both reactions enhance the quality of the “communication channel” it is rather exhausting and uncomfortable for the passengers. The situation can be improved by using in-car communication systems. These systems record the speech of each passenger by means of a single microphone or with an array of microphones. The recorded signals of the currently speaking passengers are processed by the system and played back via those loudspeakers which are located close to the non-active passengers. Comparable to public address systems, in-car communication systems operate within a closed electro-acoustic loop. Thus, signal processing is required to guarantee stable operation so as to avoid acoustic feedback such as howling or whistling. In this contribution we describe the basic processing units of an in-car communication system. Those units contain mostly standard algorithms such as beamforming, echo cancellation, and loss control. However, these methods cannot be applied and controlled as in applications like hands-free telephones or preprocessing for speech recognition systems. Here, the problem is that the excitation signals and the distorting components are highly correlated—leading to convergence problems of adaptive algorithms. Furthermore, in-car communication systems have very restrictive demands on the tolerable processing delay.

Statistical invisibility for collusion-resistant digital video watermarking

We present a theoretical framework for the linear collusion analysis of watermarked digital video sequences, and derive a new theorem equating a definition of statistical invisibility, collusion-resistance, and two practical watermark design rules. The proposed framework is simple and intuitive; the basic processing unit is the video frame and we consider second-order statistical descriptions of their temporal inter-relationships. Within this analytical setup, we define the linear frame collusion attack, the analytic notion of a statistically invisible video watermark, and show that the latter is an effective counterattack against the former. Finally, to show how the theoretical results detailed in this paper can easily be applied to the construction of collusion-resistant video watermarks, we encapsulate the analysis into two practical video watermark design rules that play a key role in the subsequent development of a novel collusion-resistant video watermarking algorithm discussed in a companion paper.

A Dynamic Architecture for Distributing Geographic Information Services

Traditional GISystems are no longer appropriate for modern distributed, heterogeneous network environments due to their closed architecture, and their lack of interoperability, reusability, and flexibility. Distributed GIServices can provide broader capabilities and functions for data management, browsing, and exchange. This paper introduces a dynamic architecture for distributing GIServices. The term dynamic indicates that the architecture is constructed temporarily by connecting or migrating geodata objects and GIS components across a network. The intention of the paper is to overview components and protocols necessary for a workable implementation of dynamically linked GIServices. The paper introduces a metadata scheme for both geodata objects and software components, and proposes an implementation framework based on existing languages, computing architectures and web services. In the framework, GIS nodes form the basic processing unit. All GIServices can be accomplished through collaboration between GIS nodes. The design of the presented framework emphasizes scalability, reusability, and dynamic integration. Current distributed computing environments cannot fully support dynamic architectures for technical and other reasons. Throughout the discussion, we distinguish what currently can be implemented from what cannot. We summarize costs and benefits of adopting a dynamic GIServices paradigm in a final section.

Lexical analysis for Chinese‐ difficulties and possible solutions†

Chinese sentences are composed with strings of characters without blanks to mark word boundaries. However, the basic processing unit for sentence processing is the word. It is the smallest meaningful, freely used unit for any natural language. Therefore lexical analysis became the first step in processing Chinese sentences. Usually a lexicon is utilized to match words and provide their syntactic and semantic information in the process of lexical analysis. During the word matching process, problems of segmentation ambiguity and occurrences of unknown words will occur. In this paper, both statistical methods and rule‐based methods are discussed for their advantages and disadvantages in solving segmentation ambiguities. For unknown word identification, off‐line word extraction methods and on‐line unknown word identification strategies are surveyed. Both methods complement each other in solving the problem. The strategies and knowledge sources for implementing a practical system are also discussed.

BALANCING STOCHASTIC KNOWLEDGE ON ACOUSTICS AND LINGUISTICS

Since basic processing units, i.e. the speech segment and the word, are different, careful balancing between acoustic and linguistic knowledge is needed in stochastic speech recognition systems. In most systems, combining acoustic and linguistic scores is controlled by a heuristic parameter, language weight. In order to achieve a robust balance, taking sequence length into account is important because the length of a speech segment is much shorter than a word duration. The length of the word sequence is not taken into account under n-gram language modeling, which is the most common stochastic language modeling for speech recognition. Due to this property, the optimal values of the language weight and word insertion penalty for balancing acoustic and linguistic probabilities is affected by the length of word sequence. To deal with this problem, a new language model is developed based on Bernoulli trial model taking the length of the word sequence into account. Not only better recognition accuracy but also robust balancing with acoustic probability compared with the normal n-gram model of the proposed method is confirmed through recognition experiments.

The Basic Information Processing (BIP) unit, mental speed and human cognitive abilities: Should the BIP R.I.P?

Lehrl and Fischer (1990, European Journal of Personality, 4, 259–286) reported findings derived from a series of tests that they claim measure a basic parameter of human information processing (BIP). Lehrl and Fischer also suggested that such indices share significant correlation with general intelligence. In the study presented here these propositions were examined along with the expectation that this measure would share significant correlation with speed of information processing parameters. The results obtained cast doubts on the usefulness of the theoretical model currently postulated by Lehrl and Fischer. Rather than reflecting some properties of a limited capacity system, the BIP would appear to be nothing more than a marker test for a known primary mental ability—reading speed.

Solving the motion planning problem by using neural networks

SUMMARYThis paper presents a new neural networks-based method to solve the motion planning problem, i.e. to construct a collision-free path for a moving object among fixed obstacles. Our ‘navigator’ basically consists of two neural networks: The first one is a modified feed-forward neural network, which is used to determine the configuration space; the moving object is modelled as a configuration point in the configuration space. The second neural network is a modified bidirectional associative memory, which is used to find a path for the configuration point through the configuration space while avoiding the configuration obstacles. The basic processing unit of the neural networks may be constructed using logic gates, including AND gates, OR gates, NOT gate and flip flops. Examples of efficient solutions to difficult motion planning problems using our proposed techniques are presented.

OR/AND neuron in modeling fuzzy set connectives

The paper introduces a neural network-based model of logical connectives. The basic processing unit consists of two types of generic OR and AND neurons structured into a three layer topology. Due to the functional integrity we will be referring to it as an OR/AND neuron. The specificity of the logical connectives is captured by the OR/AND neuron within its supervised learning. Further analysis of the connections of the neuron obtained in this way provides a better insight into the nature of the connectives applied in fuzzy sets by emphasizing their features of locality and interactivity. Afterward, we will study several architectures of neural networks comprising these neurons treated as their basic functional components. The numerical studies embrace both the structures formed by single OR/AND neurons and aimed at modeling logical connectives (including the Zimmermann-Zysno data set, 1980) and the networks representing various decision-making architectures. We will also propose a realization of a pseudo median filter in which the OR/AND neurons play an ultimate role. >

Fuzzy neural networks and neurocomputations

Abstract The paper discusses relationships between conceptual and computational platforms of fuzzy sets and neurocomputations. Fuzzy sets provide an interesting and powerful scheme of knowledge representation with its clear-cut logical properties. Neural networks with their superb learning capabilities play a primordial role in numerical processing. The synergy between these two paradigms will be studied. Two new classes of basic logic-oriented processing units (fuzzy neurons) will be introduced. The first one embraces aggregative neurons realizing OR, AND and mixed OR/AND operations. The neurons of the reference character constituting the second category of the processing units pertain to a referential character of processing that includes manipulation on binary relations of matching, difference, inclusion and dominance. The proposed architecture of logic processors implements the paradigm of distributed processing with the aid of logic-driven neurons. Various application domains (fuzzy controllers, machine learning, decision-making and distributed modelling) will be studied.

Fast algorithms and parallel structures for form factor evaluation

Two reduced complexity algorithms that enable the efficient determination of form factors are described. The preliminary complexity computations show that the new methods are at least five times faster than a previously proposed technique for practical problem sizes. Both methods trace back the special 3D visibility task inherent to the form factor calculation to 2D hidden-line elimination using special scane planes. The basic operations are planar rotations. Moreover, the proposed algorithms are highly parallel, pipelineable, and suitable for multiprocessor implementation. The basic processing unit is an ASIC-type CORDIC computing element.

Direct and Indirect Cooperation between Temporal and Parietal Networks for Invariant Visual Recognition

A new type of biologically inspired multilayered network is proposed to model the properties of the primate visual system with respect to invariant visual recognition (IVR). This model is based on 10 major neurobiological and psychological constraints. The first five constraints shape the architecture and properties of the network. 1. The network model has a Y-like double-branched multilayered architecture, with one input (the retina) and two parallel outputs, the "What" and the "Where," which model, respectively, the temporal pathway, specialized for "object" identification, and the parietal pathway specialized for "spatial" localization. 2. Four processing layers are sufficient to model the main functional steps of primate visual system that transform the retinal information into prototypes (object-centered reference frame) in the "What" branch and into an oculomotor command in the "Where" branch. 3. The distribution of receptive field sizes within and between the two functional pathways provides an appropriate tradeoff between discrimination and invariant recognition capabilities. 4. The two outputs are represented by a population coding: the ocular command is computed as a population vector in the "Where" branch and the prototypes are coded in a "semidistributed" way in the "What" branch. In the intermediate associative steps, processing units learn to associate prototypes (through feedback connections) to component features (through feedforward ones). 5. The basic processing units of the network do not model single cells but model the local neuronal circuits that combine different information flows organized in separate cortical layers. Such a biologically constrained model shows shift-invariant and size-invariant capabilities that resemble those of humans (psychological constraints): 6. During the Learning session, a set of patterns (26 capital letters and 2 geometric figures) are presented to the network: a single presentation of each pattern in one position (at the center) and with one size is sufficient to learn the corresponding prototypes (internal representations). These patterns are thus presented in widely varying new sizes and positions during the Recognition session: 7. The "What" branch of the network succeeds in immediate recognition for patterns presented in the central zone of the retina with the learned size. 8. The recognition by the "What" branch is resistant to changes in size within a limited range of variation related to the distribution of receptive field (RF) sizes in the successive processing steps of this pathway. 9. Even when ocular movements are not allowed, the recognition capabilities of the "What" branch are unaffected by changing positions around the learned one. This significant shift-invariance of the "What" branch is also related to the distribution of RF sizes. 10. When varying both sizes and locations, the "What" and the "Where" branches cooperate for recognition: the location coding in the "Where" branch can command, under the control of the "What" branch, an ocular movement efficient to reset peripheral patterns toward the central zone of the retina until successful recognition. This model results in predictions about anatomical connections and physiological interactions between temporal and parietal cortices.

Structure data sharing method of highly parallel inference engine pie

AbstractThis paper presents an efficient data management technique for the highly parallel reference engine PIE. PIE aims at the general‐purpose machine in the future knowledge processing. It is a machine which executes logic programming languages with a high speed in OR‐parallel, according to the goal‐rewriting model. In the implementation scheme considered by the authors for the goal‐rewriting model, the initial scheme is the one where the elementary processing unit is generated by a complete copy, and the sharing mechanisms are gradually adopted. The structure data sharing scheme proposed in this paper is a scheme, where the elementary processing units share only the ground instances among the structure data owned by the elementary processing units. In this paper, the structure data sharing scheme is estimated by the software simulations indicating its usefulness. The proposed scheme can reduce the degeneration time and the transfer time of the basic processing units, and improve the processing speed. On the other hand, there appears the overhead due to the lazy fetch, separating and storing of group instances, and garbage collection is required. From the detailed simulation for those aspects, we make a quantitative discussion.

Click placement and units of perceptual processing.

The relationship between experimental method and the perception hypothesis being examined is a particularly critical one. Here, earlier studies employing the click-placement method are surveyed, and the conclusions drawn there about the perceptual processes involved are discussed. A working hypothesis is developed which enables an examination, for the first time, of Ladefoged and Broadbent's [1960] original claim concerning basic processing units. Within a hypothetical framework of hierarchically organized perceptual decisions, the click-placement task is applied in a series of experiments designed to approach the question of processing units at segment, syllable and foot level. Evidence is found for the psychological reality of the syllable and foot. The phonetic basis of the foot is examined and found to be replaceable by top-down information. Between foot and syllable level, indications of differential word demarcation were found, dependent on differences in segmental structure.