A model for learning and imprinting with finite and infinite memory range

A generalization of the Bush-Mosteller learning model, in the sense of a response strength rationale, can simulate both a finite and an infinite memory range and describe learning as well as imprinting processes, the latter being characterized by the property that earlier observations enter with more weight into the system's response tendencies than later ones. The resulting difference equation for the response probabilities is no longer time invariant. Optimality properties of the model are discussed, in particular with respect to probability learning. Carnap's inductive probabilities are shown to provide a least mean square estimate for a stationary stochastic environment.     

Adiabatic solution of the Ohmic cable equation. General theory, homogeneous cylindrical fiber]

An adiabatic solution of the Ohmic cable equation is suggested, which reduces the non-stationary equation to a stationary form. The adiabatic length constant of the stationary equation is time-dependent. The adiabatic solutions for the boundary conditions that change in time linearly and exponentially were studied. In the latter case, the adiabatic length constant does not depend on time though it differs from the usual length constant. The cable input characteristics of exact and adiabatic solutions were compared in the cases of the voltage- and current-clamp, and electric field stimulation. The adiabatic and exact solutions are identical for the rising exponential stimuli. For the falling exponential stimuli, the adiabatic solution determines the exact asymptotic solution if the stimulus decays slower than the relaxation of initial conditions. It is propose to use linear and exponential ramp stimulation in electrotonic measurements.     

On the autocorrelation of one class of non-stationary random point processes

The autocorrelation of those non-stationary random point processes that can be transformed by a time transformation into a stationary process is derived. As an application, a closed form formula is obtained for the autocorrelation of those processes where the average number of points is sinusoidally modulated and the corresponding stationary process is defined by independent intervals identically distributed according to a gamma density of integer order.     

Song, sex and sensitive phases in the behavioural devebpment of birds

A number of parallels can be drawn between song learning and sexual imprinting, but how do the two processes interact during development? In zebra finches, the only species for which there are data on both song learning and sexual imprinting, an important difference between the two processes is that song learning usually occurs after the birds have sexually imprinted. Does sexual imprinting influence subsequent song tutor choice? Recent work comparing the song and sexual preferences of individual males suggests that a male does not necessarily choose a song tutor who is of the same species as he has sexually imprinted on.     

MECHANISMS OF AVIAN IMPRINTING: A REVIEW

Filial imprinting is the process through which early social preferences become restricted to a particular object or class of objects. Evidence is presented showing that filial preferences are formed not only as a result of learning through exposure to an object, but also under the influence of visual and auditory predispositions. The development of these predispositions is dependent upon certain non-specific experience. There is little evidence for an endogenously affected sensitive period for imprinting. It is more likely that the end of sensitivity is a result of the imprinting process itself. Similarly, it is now firmly established that filial and sexual preferences are reversible. Evidence suggests, however, that the first stimulus to which the young animal is exposed may exert a greater influence on filial preferences than subsequent stimuli. The learning process of imprinting is often regarded as being different from conventional associative learning. However, the imprinting object itself can function as a reinforcer. Recent studies have attempted to test predictions from an interpretation of filial imprinting as a form of associative learning. The first results suggest that ‘blocking’ may occur in imprinting, whilst there is no evidence for ‘overshadowing’. Social interactions with siblings and parent(-surrogates) have been shown to affect the formation of filial and sexual preferences. The influence of these interactions is particularly prominent in sexual imprinting, making earlier claims about naive species-specific biases unlikely. Although auditory stimuli play an important role in the formation of social attachments, there is little evidence for auditory imprinting per se. Auditory preferences formed as a result of mere (pre- or postnatal) exposure are relatively weak and short-lasting. Exposure to visual stimuli during auditory training significantly improves auditory learning, possibly through a process of reinforcement. It is becoming increasingly clear that filial and sexual imprinting are two different (although perhaps analogous) processes. Different mechanisms are likely to underlie the two processes, although there is evidence to suggest that the same brain region is involved in recognition of familiar stimuli in both filial and sexual imprinting. There is little evidence for a direct role of hormones in the learning process of imprinting. Androgen metabolism may be a factor constraining the development of a predisposition in the chick. Research into the neural mechanism of filial imprinting in the chick has revealed that a restricted part of the forebrain (IMHV) is likely to be a site of memory storage. Changes in synapse morphology and in the number of NMDA receptors have been found, limited to this region, and correlated with the strength of preference.     

The evolution of learning

Most processes or forms of learning have been treated almost as special creations, each as an independent process unrelated to others. This review offers an evolutionary cladogram linking nearly one hundred forms of learning and showing the paths through which they evolved. Many processes have multiple forms. There are at least five imprinting processes, eleven varieties of Pavlovian conditioning, ten of instrumental conditioning, and eight forms of mimicry and imitation. Song learning evolved independently in at least six groups of animals, and movement imitation in three (great apes, cetaceans and psittacine birds). The cladogram also involves at least eight new processes: abstract concept formation, percussive mimicry, cross-modal imitation, apo-conditioning, hybrid conditioning, proto-pantomime, prosodic mimicry, and image-mediated learning. At least eight of the processes evolved from more than one source. Multiple sources are of course consistent with modern evolutionary theory, as seen in some obligate symbionts, and gene-swapping organisms. Song learning is believed to have evolved from two processes: auditory imprinting and skill learning. Many single words evolved from three sources: vocal mimicry, discrimination learning, and abstract concept formation.     

The role of learning in fish behaviour

Summary The behavioural patterns of fish are the result of innate (built-in) patterns of maturation (developmental changes) and of learning processes (imprinting and trial-and-error learning). Innate behavioural patterns are considered to be hard-wired and inflexible. However, through learning, fish can adapt to environmental change. For instance, the homing behaviour of fish may be partly the result of the development of specific parts of the brain and partly because of changes in behaviour with experience. Similarly, one can assume that the feeding mode of fish involving snap-responses is innate, but learning enables fish to modify their foraging behaviour in response to a fluctuating environment. By reviewing these and other examples, such as the role of recognition learning and socially transmitted behaviour, one can illustrate the importance of learning in the everyday life of fishes. Although learning plays a large role in the behaviour of fishes, the learning capacity of fishes may also be useful to fisheries research and hatchery operations.     

Some remarks on the mathematical theory of nutrition of fishes

V. S. Ivlev [Experimental Ecology of Nutrition of Fishes, 1955, Moscow (in Russia)] has shown that the food uptake by fishes during a fixed interval of time is an exponential function of the concentration of food. Ivlev's equation is derived here, and it is shown that it can hold only for non-stationary conditions, such as prevailed in Ivlev's experiments. For a stationary state, the rate of food uptake should tend asymptotically to a limiting value as the concentration increases, but the variation is not exponential. Different other aspects of the problem are investigated, and definite new experimental procedures suggested. The implications of Ivlev's findings on the effect of non-uniformity of food distribution upon the rate of food consumption are studied from a mathematical point of view. The conclusion is reached that whereas a fish does not, in the process of eating, move directly to an individual food particle which it perceives, it does move more or less directly to large aggregates of particles, if the latter are distributed nonuniformly.     

Human temporal behavior and discrimination-reversal learning

The present study demonstrates that a relationship exists between individual differences in temporal behavior and individual differences in human discrimination-reversal learning behavior. When the results of performance on a time-estimation task employing the method of reproduction are compared with the results of acquisition performance on a complex form of discrimination-reversal learning task is can be demonstrated that underestimation of time is associated with faster learning and overestimation of time is associated with slower learning of the discrimination task. The experimental design was based on the historical fact that the Sechenov-Pavlov and Spence-Hull formulations assigned a primary role in learning to excitation and inhibition as intervening variables between the input of stimulation and output of response. The present study also used the tasks of time estimation and discrimination learning as dependent variables; and the concept that underestimation of time is associated with and underlying predominance of excitatory processes as a hypothetical construct. The conclusion was reached that one of the many theoretical processes which may be used to explain discrimination learning is the concept that acquisition of the correct response may be viewed as a function of the individual rates at which excitatory processes come to be conditioned to a predominance over existing inhibitory processes.     

Process Monitoring Strategies for Surface Mount Manufacturing Processes

Establishing reliable surface mount assemblies requires robust design and assembly practices, including stringent process control schemes for achieving high yield processes and high quality solder interconnects. Conventional Shewhart-based process control charts prevalent in today's complex surface mount manufacturing processes are found to be inadequate as a result of autocorrelation, high false alarm probability, and inability to detect process deterioration. Hence, new strategies are needed to circumvent the shortcomings of traditional process control techniques. In this article, the adequacy of Shewhart models in a surface mount manufacturing environment is examined and some alternative solutions and strategies for process monitoring are discussed. For modeling solder paste deposition process data, a time series analysis based on neural network models is highly desirable for both controllability and predictability. In particular, neural networks can be trained to model the autocorrelated time series, learn historical process behavior, and forecast future process performance with low prediction errors. This forecasting ability is especially useful for early detection of solder paste deterioration, so that timely remedial actions can be taken, minimizing the impact on subsequent yields of downstream processes. As for the automated component placement process where very low fraction nonconforming frequently occurs, control-charting schemes based on cumulative counts of conforming items produced prior to detection of nonconforming items is more sensitive in flagging process deterioration. For the reflow soldering and wave-soldering processes, the use of demerit control charts is appealing as it provides not only better control when various defects with a different degree of severity are encountered, but also leads to an improved ARL performance. Illustrative examples of actual process data are presented to demonstrate these approaches.     

A Variance Components Model for Analyzing Repeated Measures Designs with Non-Stationary Error Structure

When data are collected in the form of multiple measurements on several subjects, they are often analyzed as repeated measures data with some stationary error structure assumed for the errors. For data with non-stationary error structure, the multivariate model is often used. The multivariate model imposes restrictions that are often not met in practice by data of such type. At the same time, they ignore valuable information in the data that are related to time dependencies and time relations. In this paper, we propose a model that is a reparametrization of the multivariate model and is suitable to analyze general repeated measures designs with non-stationary error structure. The model is shown to be a variance components model whose components are estimated using the method of maximum likelihood. Several other properties of the model are derived and discussed including tests of significance. Finally, an example on neurological data is included to demonstrate its application in biological sciences.     

Bayesian approximations and extensions: Optimal decisions for small brains and possibly big ones too

We compared the performance of Bayesian learning strategies and approximations to such strategies, which are far less computationally demanding, in a setting requiring individuals to make binary decisions based on experience. Extending Bayesian updating schemes, we compared the different strategies while allowing for various implementations of memory and knowledge about the environment. The dynamics of the observable variables was modeled through basic probability distributions and convolution. This theoretical framework was applied to the problem of male fruit flies who have to decide which females they should court. Computer simulations indicated that, for most parameter values, approximations to the Bayesian strategy performed as well as the full Bayesian one. The linear approximation, reminiscent of the linear operator, was notably successful, and, without innate knowledge, the only successful learning strategy. Besides being less demanding in computation and thus realistic for small brains, the linear approximation was also successful at limited memory, which would translate into robustness in rapidly changing environments. Knowledge about the environment boosted the performance of the various learning strategies with maximal performance at large utilization of memory. Only for limited memory capacities, intermediate knowledge was most successful. We conclude that many animals may rely on algorithms that involve approximations rather than full Bayesian calculations because such approximations achieve high levels of performance with only a fraction of the computational requirements, in particular for extensions of Bayesian updating schemes, which can represent universal and realistic environments.     

Exploring Age-Related Changes in Dynamical Non-Stationarity in Electroencephalographic Signals during Early Adolescence

Dynamics of brain signals such as electroencephalogram (EEG) can be characterized as a sequence of quasi-stable patterns. Such patterns in the brain signals can be associated with coordinated neural oscillations, which can be modeled by non-linear systems. Further, these patterns can be quantified through dynamical non-stationarity based on detection of qualitative changes in the state of the systems underlying the observed brain signals. This study explored age-related changes in dynamical non-stationarity of the brain signals recorded at rest, longitudinally with 128-channel EEG during early adolescence (10 to 13 years of age, 56 participants). Dynamical non-stationarity was analyzed based on segmentation of the time series with subsequent grouping of the segments into clusters with similar dynamics. Age-related changes in dynamical non-stationarity were described in terms of the number of stationary states and the duration of the stationary segments. We found that the EEG signal became more non-stationary with age. Specifically, the number of states increased whereas the mean duration of the stationary segment decreased with age. These two effects had global and parieto-occipital distribution, respectively, with the later effect being most dominant in the alpha (around 10 Hz) frequency band.     

Mitotic index, influx and mean transit time in the hamster cheek pouch epithelium, a partially synchronized cell system. Presentation of a mathematical model based on a non-stationary probability density function for the transit time in a compartment

Mitotic activity was followed in the epithelium of the hamster cheek pouch for about 12 hr in two experiments under different conditions of noise and light intensity. A mathematical model based on a non-stationary probability density function for the transit time through mitosis was developed, making an analysis of this partially synchronized cell system possible. The most important result of the two experiments is the indication of a non-stationary mean transit time for cells in mitosis. In the first experiment (70 dB, 103 lux), which produced high mitotic indices (maximum 2.0%), the influx changed from 0. to 0.9%/hr (mean of 7-hourly determinations: 0.4%/hr), and the mean transit time from 1.9 to 5.5 hr. In the second experiment (70 dB, 265 lux), which had a lower level of MI (maximum 1.3%), the influx changed from 0 to 1.8%/hr (mean of 13-hourly determinations: 0.4%/hr), and the mean transit time from 0.3 to 3.3 hr. It was difficult to say precisely whether the variation in influx or in mean transit time was the main factor in the increase and decrease of MI. The data could not be simulated with a stationary mean transit time. It is suggested that changes in MI due to different external circumstances are mainly the result of variation in the mean transit time.     

Evolution and adaptations of Galapagos sea-birds

With two endemic genera, 26% endemism at the specific, and 58% endemism at the subspecific level, the Galapagos sea-bird fauna is more highly endemic than that of any other archipelago. Of the four most distinct, hence probably oldest, endemics, three are probably of north Pacific origin and the fourth may be. The next most distinct group is of Humboldt Current origin, and the remainder, which are very little differentiated, are of Pacific or Caribbean origin. Special adaptations to Galapagos conditions include the loss of flight by a cormorant and the evolution of nocturnal habits by a gull, the latter probably as a result of kleptoparasitism or nest-predation by frigatebirds. As a group the Galapagos sea-birds show varied adaptive responses to the relatively ascasonal and highly unpredictable marine environment. Opportunistic and non-annual breeding regimes predominate. Unlike the Galapagos land-birds, the study of Galapagos sea-birds has thrown light not on speciation processes but on the consequences of natural selection acting on breeding ecology and associated behaviour.     

Estimation of running frequency spectra using a Kalman filter algorithm

A method is suggested for the computation of running frequency spectra from non-stationary oscillations in a long time series. The method is based on an autoregressive model where the coefficients are assumed to vary slowly. The coefficients are updated using the Kalman filter technique. The method is shown to be superior to ordinary autoregressive spectral estimation based on stationary theory in recognizing rapid changes in the frequencies of oscillations.     

Monitoring the formation of an Aureobasidium pullulans biofilm in a bead-packed reactor via flow-weighted magnetic resonance imaging

The biofilm-forming fungus, Aureobasidium pullulans DSM 2404, was grown in a bead-packed reactor. Alterations within the reactor were analysed in several cross-sectional slices by magnetic resonance imaging (MRI) with flow contrast. For the first time, biofilm accumulation could be continuously elucidated without using any contrast agents, and the non-stationary flow through the fixed-bed reactor could be visualized. The results indicate that the non-stationary flow through the biofilm reactor changes significantly due to the changing reactor morphology. Preferential flow lines arise during biofilm formation. The accumulation of the biomass was determined and compared to gravimetrical biomass data. The described technique can be used to monitor hydrodynamic transport, and to combine flow-field characteristics with morphological data for the prediction of undesirable reactor processes, e.g. clogging.     

Learning about Time: Plastic Changes and Interindividual Brain Differences

Learning the timing of rapidly changing sensory events is crucial to construct a reliable representation of the environment and to efficiently control behavior. The neurophysiological mechanisms underlying the learning of time are unknown. We used functional and structural magnetic resonance imaging to investigate neurophysiological changes and individual brain differences underlying the learning of time in the millisecond range. We found that the representation of a trained visual temporal interval was associated with functional and structural changes in a sensory-motor network including occipital, parietal, and insular cortices, plus the cerebellum. We show that both types of neurophysiological changes correlated with changes of performance accuracy and that activity and gray-matter volume of sensorimotor cortices predicted individual learning abilities. These findings represent neurophysiological evidence of functional and structural plasticity associated with the learning of time in humans and highlight the role of sensory-motor circuits in the perceptual representation of time in the millisecond range.     

A Novel Classification Algorithm Based on Incremental Semi-Supervised Support Vector Machine

For current computational intelligence techniques, a major challenge is how to learn new concepts in changing environment. Traditional learning schemes could not adequately address this problem due to a lack of dynamic data selection mechanism. In this paper, inspired by human learning process, a novel classification algorithm based on incremental semi-supervised support vector machine (SVM) is proposed. Through the analysis of prediction confidence of samples and data distribution in a changing environment, a “soft-start” approach, a data selection mechanism and a data cleaning mechanism are designed, which complete the construction of our incremental semi-supervised learning system. Noticeably, with the ingenious design procedure of our proposed algorithm, the computation complexity is reduced effectively. In addition, for the possible appearance of some new labeled samples in the learning process, a detailed analysis is also carried out. The results show that our algorithm does not rely on the model of sample distribution, has an extremely low rate of introducing wrong semi-labeled samples and can effectively make use of the unlabeled samples to enrich the knowledge system of classifier and improve the accuracy rate. Moreover, our method also has outstanding generalization performance and the ability to overcome the concept drift in a changing environment.