Symbolic connectionism in natural language disambiguation

Natural language understanding involves the simultaneous consideration of a large number of different sources of information. Traditional methods employed in language analysis have focused on developing powerful formalisms to represent syntactic or semantic structures along with rules for transforming language into these formalisms. However, they make use of only small subsets of knowledge. This article describes how to use the whole range of information through a neurosymbolic architecture which is a hybridization of a symbolic network and subsymbol vectors generated from a connectionist network. Besides initializing the symbolic network with prior knowledge, the subsymbol vectors are used to enhance the system's capability in disambiguation and provide flexibility in sentence understanding. The model captures a diversity of information including word associations, syntactic restrictions, case-role expectations, semantic rules and context. It attains highly interactive processing by representing knowledge in an associative network on which actual semantic inferences are performed. An integrated use of previously analyzed sentences in understanding is another important feature of our model. The model dynamically selects one hypothesis among multiple hypotheses. This notion is supported by three simulations which show the degree of disambiguation relies both on the amount of linguistic rules and the semantic-associative information available to support the inference processes in natural language understanding. Unlike many similar systems, our hybrid system is more sophisticated in tackling language disambiguation problems by using linguistic clues from disparate sources as well as modeling context effects into the sentence analysis. It is potentially more powerful than any systems relying on one processing paradigm     

Iron,Neuroinflammation and Neurodegeneration

Disturbance of the brain homeostasis, either directly via the formation of abnormal proteins or cerebral hypo-perfusion, or indirectly via peripheral inflammation, will activate microglia to synthesise a variety of pro-inflammatory agents which may lead to inflammation and cell death. The pro-inflammatory cytokines will induce changes in the iron proteins responsible for maintaining iron homeostasis, such that increased amounts of iron will be deposited in cells in the brain. The generation of reactive oxygen and nitrogen species, which is directly involved in the inflammatory process, can significantly affect iron metabolism via their interaction with iron-regulatory proteins (IRPs). This underlies the importance of ensuring that iron is maintained in a form that can be kept under control; hence, the elegant mechanisms which have become increasingly well understood for regulating iron homeostasis. Therapeutic approaches to minimise the toxicity of iron include N-acetyl cysteine, non-steroidal anti-inflammatory compounds and iron chelation.     

Evaluation of markov program models in virtual memory systems

A first order Markov model of program behaviour is developed from FORTRAN program instruction data. The program model is evaluated by using it to generate page references for input into a simple virtual memory operating system (VMOS) simulation model. The actual trace data are also used to drive the VMOS model. In both cases the fault probability is obtained for different replacement rules, memory sizes and page sizes. A comparison of fault probabilities is used to determine the effectiveness of the Markov program model.     

Progress in the Molecular Beam Epitaxy of HgCdTe on Large-Area Si and CdZnTe Substrates

This paper presents the progress in the molecular beam epitaxy (MBE) growth of HgCdTe on large-area Si and CdZnTe substrates at Raytheon Vision Systems. We report a very high-quality HgCdTe growth, for the first time, on an 8 cm × 8 cm CdZnTe substrate. This paper also describes the excellent HgCdTe growth repeatability on multiple 7 cm × 7 cm CdZnTe substrates. In order to study the percentage wafer area yield and its consistency from run to run, small lots of dual-band long-wave infrared/long-wave infrared triple-layer heterojunction (TLHJ) layers on 5 cm × 5 cm CdZnTe substrates and single-color double-layer heterojunction (DLHJ) layers on 6-inch Si substrates were grown and tested for cutoff wavelength uniformity and micro- and macrovoid defect density and uniformity. The results show that the entire lot of 12 DLHJ-HgCdTe layers on 6-inch Si wafers meet the testing criterion of cutoff wavelength within the range 4.76 ± 0.1 μm at 130 K and micro- and macrovoid defect density of ≤50 cm⁻² and 5 cm⁻², respectively. Likewise, five out of six dual-band TLHJ-HgCdTe layers on 5 cm × 5 cm CdZnTe substrates meet the testing criterion of cutoff wavelength within the range 6.3 ± 0.1 μm at 300 K and micro- and macrovoid defect density of ≤2000 cm⁻² and 500 cm⁻², respectively, on the entire wafer area. Overall we have found that scaling our HgCdTe MBE process to a 10-inch MBE system has provided significant benefits in terms of both wafer uniformity and quality.     

Modeling of static concentrator modules incorporating lambertian or v-groove rear reflectors

Ray tracing modeling has been used to calculate the performance of static concentrator modules with a geometric concentration ratio of 2 and incorporating very narrow (1–2 mm), long and bifacial cells. The modules utilize either a v-groove or a lambertian rear reflector. It is shown that the use of very narrow cells allows a performance improvement of 5% or more compared to structures incorporating wider cells, in the case of v-groove reflectors. The averaged yearly performance for both types of reflectors is found to be rather similar, with expected light collection in the range 82–86% of that of a module with 100% cell coverage. Experimental measurements on modules with lambertian reflectors are shown to be in good agreement with the results of modeling.     

Fade duration analysis on earth-space paths at Ku-band in Fukuoka, Japan

Presented are the fade duration dynamics as observed in four close stations. Fade duration statistics derived from the collected received signal during a critical period in the region are presented and analysed. The obtained results are compared with the model release by the ITU-R P.1623 recommendation. Some modifications are then proposed.     

Smart Skin-Adhesive Patches: From Design to Biomedical Applications

With the advancement of medical and digital technologies, smart skin adhesive patches have emerged as a key player for complex medical purposes. In particular, skin adhesive patches with integrated electronics have created an excellent platform for monitoring health conditions and intelligent medication. However, the efficient design of the adhesive patches is still challenging as it requires a strong combination of network structure, adhesion, physical properties, and biocompatibility. To design an assimilated device, one must have a deep knowledge of various skin adhesive patches. This article provides a comprehensive review of the recent advances in skin-adhesive patches, including hydrogel-based adhesive patches, transdermal patches, and electronic skin (E-skin) patches, for various biomedical applications such as wound healing, drug delivery, biosensing, and health monitoring. Furthermore, the key challenges, implementable strategies, and future designs that can potentially provide researchers in designing innovative multipurpose smart skin patches are discussed. These advanced approaches are promising for managing the health and fitness of patients who require regular medical care.     

Optimised relay selection for route discovery in reactive routing

On-demand routing protocols have the potential to provide scalable information delivery in large ad hoc networks. The novelty of these protocols is in their approach to route discovery, where a route is determined only when it is required by initiating a route discovery procedure. Much of the research in this area has focused on reducing the route discovery overhead when prior knowledge of the destination is available at the source or by routing through stable links. Hence, many of the protocols proposed to date still resort to flooding the network when prior knowledge about the destination is un-available. This paper proposes a novel routing protocol for ad hoc networks, called On-demand Tree-based Routing Protocol (OTRP). This protocol combines the idea of hop-by-hop routing (as used by AODV) with an efficient route discovery algorithm called Tree-based Optimised Flooding (TOF) to improve scalability of ad hoc networks when there is no prior knowledge about the destination. To achieve this in OTRP, route discovery overheads are minimised by selectively flooding the network through a limited set of nodes, referred to as branching nodes. The key factors governing the performance of OTRP are theoretically analysed and evaluated, including the number of branch nodes, location of branching nodes and number of Route REQuest (RREQ) retries. It was found that the performance of OTRP (evaluated using a variety of well-known metrics) improves as the number of branching nodes increases and the number of consumed RREQ retries is reduced. Additionally, theoretical analysis and simulation results shows that OTRP outperforms AODV, DYMO, and OLSR with reduced overheads as the number of nodes and traffic load increases.