Memory Storage Capacity

In order to make a crude estimate of a Hopfield net s pattern storage capacity, consider first what it means for a given pattern to be stable. From equation 10.7, we see that a pattern is stable if 

From this we see that the pattern is stable if the magnitude of the second term is less than 1 for all i, since, if this is so, equation 10.21 shows that the sign of remains unchanged. 

consider the quantity obtained by multiplying this second term by 

The storage capacity can now be estimated in two steps (1) First, assume for simplicity that the set of stored patterns is completely random that is, assume that for each pattern I and neuron i, the probability of having = +1 is equai to the 

From equation 10.24 we see that is 1/A times the sum of Np independent random variables, each of which is equal to either - -1 or —1. is thus 

---

Iit78] Little, W.A and G.L.Shaw, Analytic study of the memory storage capacity of a neural network, Math. Bios. 39 (1978) 281-290. 

All this requires a high speed computer with large memory storage capacity and RAM (something not possible until recently). Image aneifysis software... 

Given a flexible computer program and a large memory-storage capacity machine, what can one do In the ideal case one would like to use an internally consistent force-field which embodies all the following features ... 

Hamad I would like to make a historical comment about modularity. Historically, the mother of all modules is Chomsky s autonomy of syntax . That is where it started. It wasn t one of the criteria for modularity. The only criterion that was needed was its functional autonomy— the fact that it could be treated in MIT linguistics by MIT linguists without having to worry about memory, storage capacity, speed and so on. It could be understood on its own terms. This is the only non-arbitrary criterion for modularity that there is the rest of it is things we ve tacked on. 

Memory storage capacity Visual Visual Visual Auditory Auditory Auditory Vibrotactile (Same options as above ) Maximum amount of information of type defined by stimulus that the subject is able to recall. Stimuli usually consist of sets of symbols of varying complexity in different sets (e.g., spatially distributed lights, alphanumeric characters, words, motions, etc.) but similar amount of information per symbol in a given set. Units of bits are ideal, but not often possible if number of bits per symbol is unknown. In such cases, units are often reported as symbols, chunks, or items ... 

Three types of computer control systems are commonly used for pilot-plant instmmentation. The first is a centralized system, usually based on a minicomputer or occasionally a mainframe. These systems have large storage capacities, substantial memories, and much associated equipment. They typically control all the pilot plants in an area or faciUty. Centralized systems are economical if a large number of units are involved but are becoming less common due to their high installation and maintenance costs as well as the limitation that any failure of the central system shuts down all pilot plants involved. 

The purpose of interfacing instruments with computers is to enable raw analytical data to be collected as it is produced, then processed, stored and displayed or printed out. This may be accomplished as it is gathered, i.e. in real-time, or at some later time, i.e. post-run. Complete chromatograms or spectra can easily be stored in the main memory or RAM or transferred to disk. The immense storage capacity of mainframe computers can be used to provide large libraries of data (data banks) for future reference. 

Neural network model composed of formal neurons without the capacity of memory storage cannot be applicable to the study of information processing of real neural networks. 

Increases in processor speeds and storage capacity allowed these system to acquire and process data rapidly. Many fourth-generation systems became nodes in laboratory computer UMS networks. They communicate with host computers to receive instructions for analyses and for transferring results. Programs and values of parameters for specific analytical methods can be stored in memory and recalled by the analyst as needed. While the analyst found interaction with these systems easier, he or she became further removed from the system components and often more dependent on the vendor s software. Tailoring requirements to individual user requirements was often not viable with this approach. 

The newborn rat is a natural split-brain preparation for olfactory learning protocols it can be trained to associate an odor with a milk reward via just one nostril. If the other nostril is tested, the animal shows no preference. However, at 12 days of age or later, the two sides of the brain are connected and a learned preference occurs with either nostril open. The information is stored M/nlaterally the animal shows unilateral preference if the commissure is cut after training. The maturation of the commissure pathways occurs between 6 and 12 days of age. In summary, unilaterally represented memories remain unilateral, even after bilateral retrieval processes have developed. The mnemonic storage capacity of the brain is increased by confining memory to one side (Kucharski and Hall, 1987). 

An interesting futuristic application is in the field of molecular electronics where a one-dimensional molecular wire such as polyacetylene in combination with a suitable molecular switch, e.g. salicylideneanilines, would yield a molecular microchip whose information storage capacity would be about 10 times that of a conventional microchip. A new generation of high performance computer with memory elements of nanometre dimensions is visualized on the basis of such molecular microchips. 

If some care is not taken, a calculation may be needlessly limited by the storage capacity of central memory, disks, or tapes. 

The hard disk of a computer exploits thin magnetic metal films with a present-day data storage capacity of about 10 Gb in-2 whereas in the case of flexible tape memory systems incorporating magnetic particles the capacity is about 100 times smaller. 

For some mass spectrometers, such as TOF mass spectrometers, for reasons of data transfer speed and data storage capacity, the data acquisition system needs to accumulate data for a period of time in the summing memory, and forward the accumulated data to the data system. Each spectrum is added to the sum of the previous spectra so that a continuous summation process takes place. This type of data acquisition system is called digital signal averaging (DSA) or integrating transient recorder (ITR). Figure 3.8 illustrates the principle of mass spectrum acquisition with this type of system. 

System performance requirements (e.g., timing, memory storage, availability, and spare capacity)... 

Besides the scientific questions related to the coupling of models, the interaction of the numerical models is a big technical challenge. The transformation of data at different temporal and spatial resolutions as well as computational efficiency, memory consumption, data storage capacity, meta-data communication and code management are issues which have to be addressed. 

A technique that is increasingly popular is molecular dynamics. This enables the study of free energies and of the effects of changing temperature and pressure. This technique is notoriously computer resource-hungry but increases in storage capacity, memory and processor speed have made it more feasible and it is now possible to combine ab initio and molecular dynamics calculations. The next section is devoted to this and related topics. 

---