Output pattern

Fig. 9. High power array of phase coupled GaAs/AlGaAs lasers mounted -side down on a thermal heat sink. The tt/2 shift of the neighboring lasers is indicated by the + and — signs. The output pattern consists of two dominant peaks, each associated with the lasers of the same phase, and much weaker...

Scale/map the data to an array of numbers. Transform the real world data into numeric input and target output patterns. 

Concept A set of chemical inputs can generate a particular light output from a light-powered molecular-level device. Such output patterns correspond to various members of the logic vocabulary. Besides offering... 

Artificial neural networks (ANN) are computing tools made up of simple, interconnected processing elements called neurons. The neurons are arranged in layers. The feed-forward network consists of an input layer, one or more hidden layers, and an output layer. ANNs are known to be well suited for assimilating knowledge about complex processes if they are properly subjected to input-output patterns about the process. 

Non-ionic materials, such as ethoxylates, do not appear to alter the heat output pattern of ordinary Portland cement. 

The DOPH membrane and the ammonium salt membrane responded to taste substances in different ways. The above results suggest that taste substances can be perceived satisfactorily using various kinds of lipid materials. Furthermore, we must improve the sensing reproducibility. As a second step, we have developed a multichannel lipid membrane taste sensor. Taste substances can be discriminated by the output pattern from several lipid membranes. 

Figure 10. Output patterns for two brands of commercial aqueous drinks and the mixed solution. Drink A is shown by a solid line with another drink by a dot-dashed line. The mixed solution (2 mM HC1, SO mM NaCl, 0.2 mM quinine, 100 mM sucrose) is shown by a dashed line, which is closer to the solid line than the dot-dashed line.

The present sensor could easily discriminate between some kinds of commercial drinks such as coffee, beer and aqueous ionic drinks (Figure 11) [22], Since the standard deviations were 2 mV at maximum in this experimental condition, these three output patterns are definitely different. If the data are accumulated in the computer, any food can be easily discriminated. Furthermore, the taste quality can also be described quantitatively by the method mentioned below. In biological systems, patterns of frequency of nerve excitation may be fed into the brain, and then foods are distinguished and their tastes are recognized [4-8]. Thus, the quality control of foods becomes possible using the taste sensor, which has a mechanism of information processing similar to biological systems. 

Figure 11. Output patterns for coffee (-), beer and commercial aqueous drink (-).

Figure 12. Output patterns for eight brands of beer. The origin of the electric potential was taken to some beer Kl.

The direct transformation from the output pattern to the taste quality was performed here as one trial of expressing the actual human sensation using the output electrical pattern. A similar trial was done for evaluation of the strengths of sourness and saltiness, which will be mentioned later. These two trials depend on the utilization of simple transformation equations by extracting typical properties of output patterns. This method is effective if some data on sensory tests, using humans as a standard, can be obtained to compare with the sensor outputs. However, the expressions for the tastes of beer are obscure because they are not described by the five basic taste qualities. The purpose of the application of the taste sensor is also to express these kinds of obscure terms of human sense in scientific terms. 

Figure 14. Taste map of beer obtained from the output patterns using transformation equation (2).

A principal component analysis was also performed [20]. It was found that relative positions among different brands of beer (for example, with respect to K2 as an origin) are similar to those in Figure 14. This assures the conventional taste expressions such as "sharp touch" and "rich taste" in the taste map. Simultaneous consideration of output patterns with various methods will make it possible to describe these obscure human taste expressions using the five basic taste qualities. 

The test foods studied here were five kinds of tomatoes. When eating food, humans first masticate the food with their teeth and then taste it. Therefore, we used a mixer in place of teeth and crushed tomatoes before measuring them. The preconditions were established by keeping the electrode immersed in standard juice, i.e., commercial canned tomato juice without NaCl added, for a long period of time. The origin of the output pattern was taken under these preconditions. Standard juice was used for the reference electrical potential pattern. The standard deviations between different lots of membrane were about 3mV. The same set of the eight membranes was used throughout the measurements for all tomatoes. 

Figure 15. Output patterns for several samples of TVR-2. Different symbols denote different samples.

We tried three methods to quantify the taste of the foodstuffs. The first method is to compare output patterns between test solution and the mixed solutions by performing many measurements of various mixed solutions (Figure 10) [22], The taste of commercial aqueous drink was reproduced by blending four basic taste substances (HC1, NaCl, quinine, sucrose) so that the response pattern could get closest to that of an aqueous drink. With this attempt, the best combination of the concentrations of basic taste substances was obtained 2 mM HC1, 50 mM NaCl, 0.2 mM quinine and 100 mM sucrose. This mixed solution produced almost the same taste as the aqueous drink. 

It is very important to note that this method automatically contains the interactions between taste substances. If many measurements of various mixed solutions are made, comparison of output patterns between test solution and the mixed solutions can be easily made by adequate algorithms such as in neural networks. 

The second method to quantify the taste by the sensor may be to extract the characteristics of output patterns by adopting some algebraic functions [19, 20]. We can know the taste quality and estimate the taste strength of test solution by using the functions (Figures 14 and 18). However, it may not be easy to get such reliable, simple functions for expressing the taste strength for each taste quality. 

Fig. 13.14. The output pattern of a 48-bit code generator. An external clock operating at 50 Hz is applied. The graph shows two full periods of 64 bits, 48 information bits plus 16 start bits. The code is indicated above.

The usefulness of the of artificial neural networks as a modelling tool is apparent. A more general H-Oil product slate model can be developed by including the feed and catalyst properties. It can also easily be adapted to model the other aspects of the H-Oil process such the hydrotreating and hydrocracking reaction kinetics or coke lay down tendency in the separation units with the appropriate input and output patterns. 

A demonstration of the predictive potential (and the limitations) of kinetic simulations is provided in Fig. 3. The left column shows the experimental output pattern determined by M. J. Berry for the laser system in Table I with initial gas mixture CF3l H2 Ar= 1 1 50 torr. The high buffer gas pressure was used to enhance rotational relaxation. The central column in the figure shows the computed results obtained by solving the rate equations (1) and (2) with R-T rate constants of the form... 

Fig. 1.6. Inter (solid line) and intra (dotted line) distances of output patterns for different levels of activity in the MB. Here, the MB size was fixed to N- c 50000 and n Q was varied from 48 to 950 (left to right). The level of 113 active KCs seems to be optimal. For some activity levels the classification success also depends rather strongly on the learning rate p. (Modified from (Huerta et al. 2004)).

In the training phase we presented 8000 inputs from the input set in random order and then characterized the resulting output patterns in the eKCs. 

---