Motivation: Classical definitions of information, such as the Shannon information, are designed for open loop systems because they define information on a channel which has an input and an output. The main motivation of this paper is to present a closed loop information measure which is compatible with constructivist thinking. Design: Our information measure for a closed loop system reflects how additional sensor inputs are utilised to establish additional sensor-motor loops during learning. Our information (...) measure is based on the assumption that it is not optimal to stay reactive and that it is beneficial to become proactive through increased learning about the environment. Consequently our information measure gauges the utilisation of new sensor inputs to generate anticipatory actions. We call this information measure "predictive information" (PI). Findings: Our PI is zero if the organism uses only its reflex reactions. It grows when the organism is able to use other sensor inputs to preempt reflex reactions and is able to replace reflexes by anticipatory reactions. This has been demonstrated with a real robot that had to learn to avoid obstacles. Conclusion: PI is a new measure which is able to quantify anticipatory learning and, in contrast to the Shannon information, is calculated only at the inputs of an agent. This information measure has been successfully applied to a simple robot task but its application is neither limited to a certain task nor to a certain learning rule. (shrink)

In this paper, I discuss the relationship between bodily experiences in dreams and the sleeping, physical body. I question the popular view that dreaming is a naturally and frequently occurring real-world example of cranial envatment. This view states that dreams are functionally disembodied states: in a majority of dreams, phenomenal experience, including the phenomenology of embodied selfhood, unfolds completely independently of external and peripheral stimuli and outward movement. I advance an alternative and more empirically plausible view of dreams as weakly (...) phenomenally-functionally embodied states. The view predicts that bodily experiences in dreams can be placed on a continuum with bodily illusions in wakefulness. It also acknowledges that there is a high degree of variation across dreams and different sleep stages in the degree of causal coupling between dream imagery, sensory input, and outward motor activity. Furthermore, I use the example of movement sensations in dreams and their relation to outward muscular activity to develop a predictive processing account. I propose that movement sensations in dreams are associated with a basic and developmentally early kind of bodily self-sampling. This account, which affords a central role to active inference, can then be broadened to explain other aspects of self- and world-simulation in dreams. Dreams are world-simulations centered on the self, and important aspects of both self- and world-simulation in dreams are closely linked to bodily self-sampling, including muscular activity, illusory own-body perception, and vestibular orienting in sleep. This is consistent with cognitive accounts of dream generation, in which long-term beliefs and expectations, as well as waking concerns and memories play an important role. What I add to this picture is an emphasis on the real-body basis of dream imagery. This offers a novel perspective on the formation of dream imagery and suggests new lines of research. (shrink)

Predictive medical information is used by underwriters to assess the future risk of a claim in medically based insurance products such as health, life, and disability insurance. Medical underwriting involves the science of evaluating medical information to determine the risk for groups of individuals with various medical conditions. In disability insurance, this involves an evaluation of medical information to predict the risk of becoming disabled.Before discussing medical underwriting, an understanding of certain terms used by disability insurance (...) companies and the products that are available is required. The first is the definition of disability. The Americans with Disabilities Act defines disability as the following:A physical or mental impairment that substantially limits one or more of the major life activities; or has a record of such an impairment; or is regarded as having such an impairment even when no impairment exists, no substantial limitation results from the impairment, or the impairment is only substantially limiting because of the attitudes of others. (shrink)

To formalize the human judgment of rhythm complexity, we used five measures from information theory and algorithmic complexity to measure the complexity of 48 artificially generated rhythmic sequences. We compared these measurements to human prediction accuracy and easiness judgments obtained from a listening experiment, in which 32 participants guessed the last beat of each sequence. We also investigated the modulating effects of musical expertise and general pattern identification ability. Entropy rate and Kolmogorov complexity were correlated with prediction accuracy, and (...) highly correlated with easiness judgments. A logistic regression showed main effects of musical training, entropy rate, and Kolmogorov complexity, and an interaction between musical training and both entropy rate and Kolmogorov complexity. These results indicate that information-theoretic concepts capture some salient features of the human judgment of rhythm complexity, and they confirm the influence of musical expertise on complexity judgments. (shrink)

Tononi et al. claim that their integrated information theory of consciousness makes testable predictions. This article discusses two of the more startling predictions, which follow from the theory’s claim that conscious experiences are generated by inactive as well as active neurons. The first prediction is that a subject’s conscious experience at a time can be affected by the disabling of neurons that were already inactive at that time. The second is that even if a subject’s entire brain is “silent,” (...) meaning that all of its neurons are inactive (but not disabled), the subject can still have a conscious experience. A few authors have noted the implausibility of these predictions—which I call the disabling prediction and the silent brain prediction—but none have considered whether they are testable. In this article, I argue that they are not. In order to make this case, I first try to clarify the distinction between active, inactive (i.e. silent), and inactivated (i.e. disabled) neurons. With this clarification in place, I show that, even putting aside practical difficulties, it is impossible to set up a valid test of either the disabling prediction or the silent brain prediction. The conditions of the tests themselves are conditions under which a response from the subject could not reasonably be interpreted as evidence of consciousness or change in consciousness. (shrink)

Since its advent, predictive DNA testing has been perceived as a technology that may have considerable impact on the quality of people’s life. The decision whether or not to use this technology is up to the individual client. However, to enable well considered decision making both the negative as well as the positive freedom of the individual should be supported. In this paper, we argue that current professional and public discourse on predictive DNA-testing is lacking when it comes (...) to supporting positive freedom, because it is usually framed in terms of risk and risk management. We show how this ‘risk discourse’ steers thinking on the good life in a particular way. We go on to argue that empirical research into the actual deliberation and decision making processes of individuals and families may be used to enrich the environment of personal deliberation in three ways: (1) it points at a richer set of values that deliberators can take into account, (2) it acknowledges the shared nature of genes, and (3) it shows how one might frame decisions in a non-binary way. We argue that the public sharing and discussing of stories about personal deliberations offers valuable input for others who face similar choices: it fosters their positive freedom to shape their view of the good life in relation to DNA-diagnostics. We conclude by offering some suggestions as to how to realize such public sharing of personal stories. (shrink)

The distinction between explanation and prediction has received much attention in recent literature, but the equally important distinction between explanation and description (or between prediction and description) remains blurred. This latter distinction is particularly important in the social sciences, where probabilistic models (or theories) often play dual roles as explanatory and descriptive devices. The distinction between explanation (or prediction) and description is explicated in the present paper in terms of information theory. The explanatory (or predictive) power of a (...) probabilistic model is identified with information taken from (or transmitted by) the environment (e.g., the independent, experimentally manipulated variables), while the descriptive power of a model reflects additional information taken from (or transmitted by) the data. Although information is usually transmitted by the data in the process of estimating parameters, it turns out that the number of free parameters is not a reliable index of transmitted information. Thus, the common practice of treating parameters as degrees-of-freedom in testing probabilistic models is questionable. Finally, this information-theoretic analysis of explanation, prediction, and description suggests ways of resolving some recent controversies surrounding the pragmatic aspects of explanation and the so-called structural identity thesis. (shrink)

Medical underwriting involves the application of actuarial science by analyzing medical information to predict the future risk of a claim. The objective is that individuals with like risk are treated in a like manner so that the premium paid is proportional to the risk of future claim.

This paper describes how Tononi’s information integration theory of consciousness was used to make detailed predictions about the distribution of phenomenal states in a spiking neural network. This network had approximately 18,000 neurons and 700,000 connections and it used models of emotion and imagination to control the eye movements of a virtual robot and avoid ‘negative’ stimuli. The first stage in the analysis was the development of a formal definition of Tononi’s theory of consciousness. The network was then analysed (...) for information integration and detailed predictions were made about the distribution of consciousness for each time step of recorded activity. This work demonstrates how an artificial system can be analysed for consciousness using a particular theory and in the future this approach could be used to make predictions about the phenomenal states associated with biological systems. (shrink)

The apoptosis protein has a central role in the development and the homeostasis of an organism. Obtaining information about the subcellular localization of apoptosis protein is very helpful to understand the apoptosis mechanism and the function of this protein. Prediction of apoptosis protein’s subcellular localization is a challenging task, and currently the existing feature extraction methods mainly rely on the protein’s primary sequence. In this paper we develop a feature extraction model based on two different descriptors of evolutionary (...) class='Hi'>information, which contains the 192 frequencies of triplet codons (FTC) in the RNA sequence derived from the protein’s primary sequence and the 190 features from a detrended forward moving-average cross-correlation analysis (DFMCA) based on a position-specific scoring matrix (PSSM) generated by the PSI-BLAST program. Hence, this model is called FTC-DFMCA-PSSM. A 382-dimensional (382D) feature vector is constructed on the ZD98, ZW225 and CL317 datasets. Then a support vector machine is adopted as classifier, and the jackknife cross-validation test method is used for evaluating the accuracy. The overall prediction accuracies are further improved by an objective and rigorous jackknife test. Our model not only broadens the source of the feature information, but also provides a more accurate and reliable automated calculation method for the prediction of apoptosis protein’s subcellular localization. (shrink)

Since its advent, predictive DNA testing has been perceived as a technology that may have considerable impact on the quality of people’s life. The decision whether or not to use this technology is up to the individual client. However, to enable well considered decision making both the negative as well as the positive freedom of the individual should be supported. In this paper, we argue that current professional and public discourse on predictive DNA-testing is lacking when it comes (...) to supporting positive freedom, because it is usually framed in terms of risk and risk management. We show how this ‘risk discourse’ steers thinking on the good life in a particular way. We go on to argue that empirical research into the actual deliberation and decision making processes of individuals and families may be used to enrich the environment of personal deliberation in three ways: (1) it points at a richer set of values that deliberators can take into account, (2) it acknowledges the shared nature of genes, and (3) it shows how one might frame decisions in a non-binary way. We argue that the public sharing and discussing of stories about personal deliberations offers valuable input for others who face similar choices: it fosters their positive freedom to shape their view of the good life in relation to DNA-diagnostics. We conclude by offering some suggestions as to how to realize such public sharing of personal stories. (shrink)

In this study, the focus was on the development of green energy and future prediction for the consumption of current energy sources and green energy development using an improved deep learning algorithm. In addition to the analysis of the current energy consumption used for the natural gas and oil as fuel, deep neural network algorithm is used to train the system as well as to process the data obtained previously, ranging from literature from the year 2003 until the year 2019, (...) for consumption of fuel. Also, using the proposed algorithm to predict the development of green energy consumption till 2030 is presented in terms of solar and wind generators. The resulting study also focuses on depletion of energy currently used or pollution caused because of it. The green energy controlling issue can take effect by using multiple layers of handling different features extracted from different sources and then learning the system to control it.This study aims to take advantage of carbon emissions to reduce their impact and dependence in the future on environmentally friendly renewable energies. Predicting the correct and precise amount of energy consumption and increasing the amount of environmentally friendly energy lead to a healthy ecosystem. The expected green energy consumption in the future is almost 78.25 EJ in 2030 and will be, in total energy average, 56% in 2045. The aim is to reduce dependency on costly and environmentally harmful fuels. (shrink)

Model validation in computational cognitive psychology often relies on methods drawn from the testing of theories in experimental physics. However, applications of these methods to computational models in typical cognitive experiments can hide multiple, plausible sources of variation arising from human participants and from stochastic cognitive theories, encouraging a “model fixed, data variable” paradigm that makes it difficult to interpret model predictions and to account for individual differences. This article proposes a likelihood‐based, “data fixed, model variable” paradigm in which models (...) are treated as stochastic processes in experiments with participant‐to‐participant variation that can be applied to a broad range of mechanistic cognitive architectures. This article discusses the implementation and implications of this view in model validation, with a concrete focus on a simple class of ACT‐R models of cognition. This article is not intended as a recipe for broad application of these preliminary, proof‐of‐concept methods, but as a framework for communication between statisticians searching for interesting problems in the cognitive modeling sphere, and cognitive modelers interested in generalizing from deterministic to stochastic model validation, in the face of random variation in human experimental data. (shrink)

Human interaction has two principle functions: building and maintaining relationships with others and exchanging information. The function of building and maintaining relationships with others relates to interpersonal coordination; this behavior pattern is expected to predict the outcome of social relationships, such as between therapists and patients. It is unclear, however, whether the exchange of information is associated with interpersonal coordination. In the present study, we tested a hypothesis of whether body movement synchrony occurs in a natural conversation and (...) whether this synchrony has a positive correlation with the degree of information exchange. Fifty participants were engaged in a conversation task; each had different roles in the conversation. We measured their body movements during this conversation using an optical motion capture system. Similar to methods that can be found in previous research, we calculated body movements and quantified their synchrony applying the methods previously reported that automatically quantified their body movements. Moreover, we determined the participants’ degree of information exchange concerning the conversation using a questionnaire. We observed that the body movement synchrony of pairs who talked with each other was significantly higher than that of pairs who did not talk with each other, and that this synchrony was positively associated with the degree of information exchange. These results suggest that body movement synchrony predicted information exchange. (shrink)

Recent theory suggests that action prediction relies of a motor emulation mechanism that works by mapping observed actions onto the observer action system so that predictions can be generated using that same predictive mechanisms that underlie action control. This suggests that action prediction may be more accurate when there is a more direct mapping between the stimulus and the observer. We tested this hypothesis by comparing prediction accuracy for two stimulus types. A mannequin stimulus which contained information about (...) the effectors used to produce the action and a point stimulus, which contained identical dynamic information but no effector information. Prediction was more accurate for the mannequin stimulus. However, this effect was dependent on the observer having previous experience performing the observed action. This suggests that experienced and na¨ıve observers might generate predictions in qualitatively difference ways, which may relate to the presence of an internal representation of the action laid down through action performance. (shrink)

Road link speed is one of the important indicators for traffic states. In order to incorporate the spatiotemporal dynamics and correlation characteristics of road links into speed prediction, this paper proposes a method based on LDA and GCN. First, we construct a trajectory dataset from map-matched GPS location data of taxis. Then, we use the LDA algorithm to extract the semantic function vectors of urban zones and quantify the spatial dynamic characteristics of road links based on taxi trajectories. Finally, we (...) add semantic function vectors to the dataset and train a graph convolutional network to learn the spatial and temporal dependencies of road links. The learned model is used to predict the future speed of road links. The proposed method is compared with six baseline models on the same dataset generated by GPS equipped on taxis in Shenzhen, China, and the results show that our method has better prediction performance when semantic zoning information is added. Both composite and single-valued semantic zoning information can improve the performance of graph convolutional networks by 6.46% and 8.35%, respectively, while the baseline machine learning models work only for single-valued semantic zoning information on the experimental dataset. (shrink)

Research has established that exposure to a combination of diagnostic (i.e., relevant) and nondiagnostic (i.e., irrelevant) information results in predictions that are more regressive than predictions based on diagnostic information (Hackenbrack, 1992; Hoffman and Patton, 1997). This phenomenon has been labeled the dilution effect (e.g., Tetlock and Boettger, 1989) and has been documented when individuals make predictions. This study tests for the dilution effect when small groups make predictions, and examines the effect of using a procedure designed to (...) reduce the dilution effect. Results indicate that group predictions are influenced by nondiagnostic information in the same manner as are individual predictions, and allowing participants to rate the diagnosticity of information prior to making predictions does not reduce the dilution effect. (shrink)

Emotional Intelligence (EI) has been conceptualized in the literature either as a dispositional tendency, in line with a personality trait (trait EI; Petrides and Furnham, 2001), or as an ability, moderately correlated with general intelligence (ability EI; Mayer and Salovey, 1997). Surprisingly, there have been few empirical attempts conceptualizing how the different EI approaches should be related to each other. However, understanding how the different approaches of EI may be interwoven and/or complementary is of primary importance for clarifying the conceptualization (...) of EI and organizing the literature around it. We introduce a theoretical framework explaining how trait EI, ability EI, and emotion information processing – a novel component related to EI recently introduced in the literature (e.g., Fiori and Vesely Maillefer, 2018) – may contribute to effective emotion-related performance and provide initial evidence supporting its usefulness in predicting EI-related outcomes. More specifically, we show that performance in a task in which participants had to infer the mental and emotional states of others, namely a Theory of Mind task, was predicted jointly (e.g., interaction effects) by trait EI, ability EI, and emotion information processing, after controlling for personality and IQ (N = 323). Our results argue for the importance of investigating the joint contribution of different aspects of EI in explaining variability in emotionally laden outcomes. (shrink)

The early right anterior negativity (ERAN) in event-related potentials (ERPs) is typically elicited by syntactically unexpected events in Western tonal music. We examined how visual predictive information influences syntactic processing, how musical or non-musical cues have different effects, and how they interact with sequential effects between trials, which could modulate with the strength of the sense of established tonality. The EEG was recorded from musicians who listened to chord sequences paired with one of four types of visual stimuli; (...) two provided predictive information about the syntactic validity of the last chord through either musical notation of the whole sequence, or the word “regular” or “irregular”, while the other two, empty musical staves or a blank screen, provided no information. Half of the sequences ended with the syntactically invalid Neapolitan sixth chord, while the other half ended with the Tonic chord. Clear ERAN was observed in frontocentral electrodes in all conditions. A principal component analysis was performed on the grand average response in the audio-only condition, to separate spatio-temporal dynamics of different scalp areas as principal components and use them to extract auditory-related neural activities in the other visual-cue conditions. The first principal component (PC1) showed a symmetrical frontocentral topography, while the second (PC2) showed a right-lateralized frontal concentration. A source analysis confirmed the relative contribution of temporal sources to the former and a right frontal source to the latter. Cue predictability affected only the ERAN projected onto PC1, especially when the previous trial ended with the Tonic chord. The ERAN in PC2 was reduced in the trials following Neapolitan endings in general. However, the extent of this reduction differed between cue-types, whereby it was nearly absent when musical notation was used, regardless of whether the staves were filled with notes or empty. The results suggest that the right frontal areas carry out the primary role in musical syntactic analysis and integration of the ongoing context, which produce schematic expectations that, together with the veridical expectation incorporated by the temporal areas, inform musical syntactic processing in musicians. (shrink)

Predictive information is a popular and promising family of information-based theories of biological communication. It is difficult to adjudicate between predictive information-based theories and influence-based theories of biological communication because the same acts seem to count as communicative on both theories. In this paper, I argue that predictive information theories and influence-based theories give importantly different descriptions of deceptive signals in some non-evolutionarily stable communicative systems by citing a novel case observed in nature. (...) Moreover, predictive information gives a counter-intuitive description to the case while some of its rival influence-based theories do not. I argue that there are no clear ways for defenders of predictive information to respond to this apparent problem without sacrificing important virtues of their theory or deflating the difference between the rival views. (shrink)

Predictive information is a popular and promising family of information-based theories of biological communication. It is difficult to adjudicate between predictive information-based theories and influence-based theories of biological communication because the same acts seem to count as communicative on both theories. In this paper, I argue that predictive information theories and influence-based theories give importantly different descriptions of deceptive signals in some non-evolutionarily stable communicative systems by citing a novel case observed in nature. (...) Moreover, predictive information gives a counter-intuitive description to the case while some of its rival influence-based theories do not. I argue that there are no clear ways for defenders of predictive information to respond to this apparent problem without sacrificing important virtues of their theory or deflating the difference between the rival views. (shrink)

Research has established that exposure to a combination of diagnostic (i.e., relevant) and nondiagnostic (i.e., irrelevant) information results in predictions that are more regressive than predictions based on diagnostic information (Hackenbrack, 1992; Hoffman and Patton, 1997). This phenomenon has been labeled the dilution effect (e.g., Tetlock and Boettger, 1989) and has been documented when individuals make predictions. This study tests for the dilution effect when small groups make predictions, and examines the effect of using a procedure designed to (...) reduce the dilution effect. Results indicate that group predictions are influenced by nondiagnostic information in the same manner as are individual predictions, and allowing participants to rate the diagnosticity of information prior to making predictions does not reduce the dilution effect. (shrink)

Journal Name: Cognitive Linguistics Issue: Ahead of print.

The American justice system, from police departments to the courts, is increasingly turning to information technology for help identifying potential offenders, determining where, geographically, to allocate enforcement resources, assessing flight risk and the potential for recidivism amongst arrestees, and making other judgments about when, where, and how to manage crime. In particular, there is a focus on machine learning and other data analytics tools, which promise to accurately predict where crime will occur and who will perpetrate it. Activists and (...) academics have begun to raise critical questions about the use of these tools in policing contexts. In this chapter, I review the emerging critical literature on predictive policing and contribute to it by raising ethical questions about the use of predictive analytics tools to identify potential offenders. Drawing from work on the ethics of profiling, I argue that the much-lauded move from reactive to preemptive policing can mean wrongfully generalizing about individuals, making harmful assumptions about them, instrumentalizing them, and failing to respect them as full ethical persons. I suggest that these problems stem both from the nature of predictive policing tools and from the sociotechnical contexts in which they are implemented... (shrink)

Computer abuse by employees is a critical concern for managers. Misuse of an organization’s information assets leads to costly damage to an organization’s reputation, decreases in sales, and impositions of fines. We use this opportunity to introduce and expand the theoretic framework proffered by Thong and Yap to better understand the factors that lead individuals to commit CA in organizations. The study uses a survey of 449 respondents from the banking, financial, and insurance industries. Our results indicate that individuals (...) who adhere to a formalist ethical perspective are significantly less likely to engage in CA activities than those following a utilitarian ethical framework. In addition, the results provide evidence that employees with individualistic natures are linked to increased CA incidents, whereas collectivist tendencies are associated with decreases in CA behaviors. Our results also show that collectivism acts as a strong moderator that further decreases the relationships between formalism and CA, and utilitarianism and CA. Finally, we offer detailed suggestions on how organizations and researchers can leverage our findings to decrease CA occurrences. (shrink)

The comprehension of anaphoric relations may be guided not only by discourse, but also syntactic information. In the literature on online processing, however, the focus has been on audible pronouns and descriptions whose reference is resolved mainly on the former. This paper examines one relation that both lacks overt exponence, and relies almost exclusively on syntax for its resolution: adjunct control, or the dependency between the null subject of a non-finite adjunct and its antecedent in sentences such as Mickey (...) talked to Minnie before ___ eating. Using visual world eyetracking, we compare the time course of interpreting this null subject and overt pronouns (Mickey talked to Minnie before he ate). We show that when control structures are highly frequent, listeners are just as quick to resolve reference in either case. When control structures are less frequent, reference resolution based on structural information still occurs upon hearing the non-finite verb, but more slowly, especially when unaided by structural and referential predictions. This may be due to increased difficulty in recognizing that a referential dependency is necessary. These results indicate that in at least some contexts, referential expressions whose resolution depends on very different sources of information can be resolved approximately equally rapidly, and that the speed of interpretation is largely independent of whether or not the dependency is cued by an overt referring expression. (shrink)

The central idea behind this paper is that presuppositions of soft triggers arise from the way our attention structures the informational content of a sentence. Some aspects of the information conveyed are such that we pay attention to them by default, even in the absence of contextual information. On the other hand, contextual cues or conversational goals can divert attention to types of information that we would not pay attention to by default. Either way, whatever we do (...) not pay attention to, be it by default, or in context, is what ends up presupposed by soft triggers. This paper attempts to predict what information in the sentence is likely to end up being the main point (i.e. what we pay attention to) and what information is independent from this, and therefore likely presupposed. It is proposed that this can be calculated by making reference to event times. The notion of aboutness used to calculate independence is based on that of Demolombe and Fariñas del Cerro (In: Holdobler S (ed) Intellectics and computational logic: papers in honor of Wolfgang Bibel, 2000). (shrink)

Statistics constitute the social universe of which they are gathered. The foundation necessary to develop quantified knowledge about society is the population. If quantified knowledge changes society, the question arises on how individuals become to be represented as population. The population has to be extracted from individuals in a process that we call “populationisation.” This encompasses the development of the individual into a segment of a population through the compilation of individual data into population data and its analysis. To describe (...) the process of populationisation, we follow a statistical risk assessment tool, the German Diabetes Risk Score, from its inception in a cohort study to its entry into the public sphere of German society. The population is extracted from individuals that function as research subjects through an identification number. Preprocessed information is entered into a dynamic database that enables its rearrangement according to main aggregates. Through populationisation, a signified is enacted that can be sliced up and that is equal in all its parts. Now predictions can be computed and fed back to society as tools of moral judgment. The acting individual and “society” are—numerically and literally!—entangled in the production and enactment of risk knowledge. (shrink)

In this paper an Artificial Neural Network (ANN) model, for predicting the category of a tumor was developed and tested. Taking patients’ tests, a number of information gained that influence the classification of the tumor. Such information as age, sex, histologic-type, degree-of-diffe, status of bone, bone-marrow, lung, pleura, peritoneum, liver, brain, skin, neck, supraclavicular, axillar, mediastinum, and abdominal. They were used as input variables for the ANN model. A model based on the Multilayer Perceptron Topology was established and (...) trained using data set which its title is “primary tumor” and was obtained from the University Medical Centre, Institute of Oncology, Ljubljana, Yugoslavia Test data evaluation shows that the ANN model is able to correctly predict the tumor category with 76.67 % accuracy. (shrink)