Yury Shimansky

About
Artificial Intelligence Neuroscience Mathematics Stock Trading
Publications Resume Contact

The list of publications below is shown by categories. Please use the links on the left to navigate. Move cursor over [abstract] to view paper abstract.

Artificial intelligence

Shimansky Y.P. (2020) Trans-algorithmic constructiveness of life emergence and evolution. Biological Cybernetics. (submitted) [abstract]
A general theory of life emergence is highly desirable for understanding the essence of life. Here a conceptual framework of trans-algorithmic (TA) universal learning systems (ULS) is introduced and its capacity for describing what biosystems are, their emergence, and evolution is demonstrated. It is proposed that life begins with emergence of systems that can be described as being stable due to performance of a special function. An example of a physical system with function-based stability is provided, and it is suggested that such systems can be considered as proto-biosystems. Biosystems are described as possessing an optimized control subsystem as an additional level of the functional hierarchy. Life evolution is described in terms of complexity increases that improve function-based stability of biosystems as TA ULS. It is essentially a TA process. In this regard, for example, addition of new sensors and effectors is viewed as useful increases in the observability and controllability of the controlled object. It is demonstrated that emergence of consciousness can be presented in TA ULS terms as a natural, intuitively clear outcome of the general tendency to increase useful complexity in populations of closely cooperating biosystems. Deep roots of TA ULS concept in both theoretical physics, computation theory, and cybernetics are revealed. A comparison between TA ULS framework and existing approaches to describing life and its emergence is discussed. In conclusion, the conceptual framework of TA ULS is a unifying theory that complements the existing mainstream approaches to life description rather than contesting them.

Shimansky Y.P. (2018) Trans-algorithmic nature of learning in biological systems. Biological Cybernetics. 112(4):357-368. [abstract]
Learning ability is a vitally important, distinctive property of biological systems, which provides dynamic stability in nonstationary environments. Although several different types of learning have been successfully modeled using a universal computer, in general, learning cannot be described by an algorithm. In other words, algorithmic approach to describing the functioning of biological systems is not sufficient for adequate grasping of what is life. Since biosystems are parts of the physical world, one might hope that adding some physical mechanisms and principles to the concept of algorithm could provide extra possibilities for describing learning in its full generality. However, a straightforward approach to that through the so-called physical hypercomputation so far has not been successful. Here an alternative approach is proposed. Biosystems are described as achieving enumeration of possible physical compositions though random incremental modifications inflicted on them by active operating resources (AORs) in the environment. Biosystems learn through algorithmic regulation of the intensity of the above modifications according to a specific optimality criterion. From the perspective of external observers, biosystems move in the space of different algorithms driven by random modifications imposed by the environmental AORs. A particular algorithm is only a snapshot of that motion, while the motion itself is essentially trans-algorithmic. In this conceptual framework, death of unfit members of a population, for example, is viewed as a trans-algorithmic modification made in the population as a biosystem by environmental AORs. Numerous examples of AOR utilization in biosystems of different complexity, from viruses to multicellular organisms, are provided.

Shimansky Y.P. (2011) State estimation bias induced by optimization under uncertainty and error cost asymmetry is likely reflected in perception. Biological Cybernetics. 104(4): 225-233. [abstract]
It is well known from numerous studies that perception can be significantly affected by intended action in many everyday situations, indicating that perception and related decision-making is not a simple, one-way sequence, but a complex iterative cognitive process. However, the underlying functional mechanisms are yet unclear. Based on an optimality approach, a quantitative computational model of one such mechanism has been developed in this study. It is assumed in the model that significant uncertainty about task-related parameters of the environment results in parameter estimation errors and an optimal control system should minimize the cost of such errors in terms of the optimality criterion. It is demonstrated that, if the cost of a parameter estimation error is significantly asymmetrical with respect to error direction, the tendency to minimize error cost creates a systematic deviation of the optimal parameter estimate from its maximum-likelihood value. Consequently, optimization of parameter estimate and optimization of control action cannot be performed separately from each other under parameter uncertainty combined with asymmetry of estimation error cost, thus making the certainty equivalence principle non-applicable under those conditions. A hypothesis that not only the action, but also perception itself is biased by the above deviation of parameter estimate is supported by ample experimental evidence. The results provide important insights into the cognitive mechanisms of interaction between sensory perception and planning an action under realistic conditions. Implications for understanding related functional mechanisms of optimal control in the CNS are discussed.

Shimansky Y.P. (2010) Adaptive force produced by stress-induced regulation of random variation intensity. Biological Cybernetics. 103(2): 135-150. [abstract]
The Darwinian theory of life evolution is capable of explaining the majority of related phenomena. At the same time, the mechanisms of optimizing traits beneficial to a population as a whole but not directly to an individual remain largely unclear. There are also significant problems with explaining the phenomenon of punctuated equilibrium. From another perspective, multiple mechanisms for the regulation of the rate of genetic mutations according to the environmental stress have been discovered, but their precise functional role is not well understood yet. Here a novel mathematical paradigm called a Kinetic-Force Principle (KFP), which can serve as a general basis for biologically plausible optimization methods, is introduced and its rigorous derivation is provided. Based on this principle, it is shown that, if the rate of random changes in a biological system is proportional, even only roughly, to the amount of environmental stress, a virtual force is created, acting in the direction of stress relief. It is demonstrated that KFP can provide important insights into solving the above problems. Evidence is presented in support of a hypothesis that the nature employs KFP for accelerating adaptation in biological systems. A detailed comparison between KFP and the principle of variation and natural selection is presented and their complementarity is revealed. It is concluded that KFP is not a competing alternative, but a powerful addition to the principle of variation and natural selection. It is also shown KFP can be used in multiple ways for adaptation of individual biological organisms.

Shimansky Y.P. (2009) Biologically plausible learning in neural networks: a lesson from bacterial chemotaxis. Biological Cybernetics. 101(5): 379-385. [abstract]
Learning processes in the brain are usually associated with plastic changes made to optimize the strength of connections between neurons. Although many details related to biophysical mechanisms of synaptic plasticity have been discovered, it is unclear how the concurrent performance of adaptive modifications in a huge number of spatial locations is organized to minimize a given objective function. Since direct experimental observation of even a relatively small subset of such changes is not feasible, computational modeling is an indispensable investigation tool for solving this problem. However, the conventional method of error back-propagation (EBP) employed for optimizing synaptic weights in artificial neural networks is not biologically plausible. This study based on computational experiments demonstrated that such optimization can be performed rather efficiently using the same general method that bacteria employ for moving closer to an attractant or away from a repellent. With regard to neural network optimization, this method consists of regulating the probability of an abrupt change in the direction of synaptic weight modification according to the temporal gradient of the objective function. Neural networks utilizing this method (regulation of modification probability, RMP) can be viewed as analogous to swimming in the multidimensional space of their parameters in the flow of biochemical agents carrying information about the optimality criterion. The efficiency of RMP is comparable to that of EBP, while RMP has several important advantages. Since the biological plausibility of RMP is beyond a reasonable doubt, the RMP concept provides a constructive framework for the experimental analysis of learning in natural neural networks.

Shimansky Y.P. (2007) Role of optimization in simple and learning-based adaptation and its biologically plausible mechanisms. In: T.O. Williams (ed.) Biological cybernetics research trends. Nova Science Publishers: NY, pp. 95-164. [abstract]
Adaptation and learning in biological systems are usually modeled based on optimization of parameters and structure with respect to a specific criterion. However, the adequacy of the optimization approach to describing the behavior of a biosystem is not always obvious. In addition, optimization methods utilized in computer models often are not biologically feasible, which obstructs comparison between experimental and modeling results and hinders our understanding of biological optimization mechanisms. This study is focused on addressing these problems and creating a theoretical basis for the development of biologically plausible optimization methods. A novel mathematical paradigm called here the kinetic-force principle (KFP), which can serve as a basis for biologically plausible optimization methods applicable to any system, is derived and its functioning is described on several computer models. Ample evidence is presented in support of a hypothesis that the nature employs KFP in many applications, particularly in addition to natural selection, for accelerating evolutionary adaptation based on the regulation of the rate of inheritable random changes depending of the amount of environmental stress. A computational approach to modeling functional mechanisms of learning is used to analyze the functional organization of the internal memory structure that supports learning processes in biosystems. The concept of a universal learning system (ULS) introduced by the author in previous studies is developed further. The functional architecture of a ULS is explored in detail, emphasizing the application of fundamental results obtained in the reinforcement learning (RL) theory for computational modeling in biology and neuroscience, with a focus on motor learning. It is demonstrated how the RL method for online solution of an optimal control problem of infinite-horizon type can be combined with the optimal path leaning method, recently developed by the author, to form a functionally powerful hierarchical learning system. It is shown that the combined system is capable of solving such important problems as instrumental reflex development much more efficiently than the above RL method alone and in much better correspondence with experimental observations.

Shimansky, Y.P. (2004) The concept of a Universal Learning System as a basis for creating a general mathematical theory of learning. Minds and Machines. 14(4): 453-484. [abstract]
The number of studies related to natural and artificial mechanisms of learning rapidly increases. However, there is no general theory of learning that could provide a unifying basis for exploring different directions in this growing field. For a long time the development of such a theory has been hindered by nativists’ belief that the development of a biological organism during ontogeny should be viewed as parameterization of an innate, encoded in the genome structure by an innate algorithm, and nothing essentially new is created during this process. Noam Chomsky has claimed, therefore, that the creation of a non-trivial general mathematical theory of learning is not feasible, since any algorithm cannot produce a more complex algorithm. This study refutes the above argumentation by developing a counter-example based on the mathematical theory of algorithms and computable functions. It introduces a novel concept of a Universal Learning System (ULS) capable of learning to control in an optimal way any given constructive system from a certain class. The necessary conditions for the existence of a ULS and its main functional properties are investigated. The impossibility of building an algorithmic ULS for a sufficiently complex class of controlled objects is shown, and a proof of the existence of a nonalgorithmic ULS based on the axioms of classical mathematics is presented. It is argued that a non-algorithmic ULS is a legitimate object of not only mathematics, but also the world of nature. These results indicate that an algorithmic description of the organization and adaptive development of biological systems in general is not sufficient. At the same time, it is possible to create a rigorous non-algorithmic general theory of learning as a theory of ULS. The utilization of this framework for integrating learning-related studies is discussed.

Shimansky, Y.P., Kang, T., and He, J. (2004) A novel model of motor learning capable of developing an optimal movement trajectory on-line from scratch. Biological Cybernetics. 90(2): 133-45. [abstract]
A computational model of a learning system (LS) is described that acquires knowledge and skill necessary for optimal control of a multisegmental limb dynamics (controlled object or CO), starting from ‘‘knowing’’ only the dimensionality of the object’s state space. It is based on an optimal control problem setup different from that of reinforcement learning. The LS solves the optimal control problem online while practicing the manipulation of CO. The system’s functional architecture comprises several adaptive components, each of which incorporates a number of mapping functions approximated based on artificial neural nets. Besides the internal model of the CO’s dynamics and adaptive controller that computes the control law, the LS includes a new type of internal model, the minimal cost (IMmc) of moving the controlled object between a pair of states. That internal model appears critical for the LS’s capacity to develop an optimal movement trajectory. The IMmc interacts with the adaptive controller in a cooperative manner. The controller provides an initial approximation of an optimal control action, which is further optimized in real time based on the IMmc. The IMmc in turn provides information for updating the controller. The LS’s performance was tested on the task of center-out reaching to eight randomly selected targets with a 2DOF limb model. The LS reached an optimal level of performance in a few tens of trials. It also quickly adapted to movement perturbations produced by two different types of external force field. The results suggest that the proposed design of a selfoptimized control system can serve as a basis for the modeling of motor learning that includes the formation and adaptive modification of the plan of a goal-directed movement.

Shimansky, Y.P. (2000) Spinal motor control system incorporates an internal model of limb dynamics. Biological Cybernetics. 83: 379-389. [abstract]
The existence and utilization of an internal representation of the controlled object is one of the most important features of the functioning of neural motor control systems. This study demonstrates that this property already exists at the level of the spinal motor control system (SMCS), which is capable of generating motor patterns for reflex rhythmic movements, such as locomotion and scratching, without the aid of the peripheral afferent feedback, but substantially modifies the generated activity in response to peripheral afferent stimuli. The SMCS is presented as an optimal control system whose optimality requires that it incorporate an internal model (IM) of the controlled object's dynamics. A novel functional mechanism for the integration of peripheral sensory signals with the corresponding predictive output from the IM, the summation of information precision (SIP) is proposed. In contrast to other models in which the correction of the internal representation of the controlled object's state is based on the calculation of a mismatch between the internal and external information sources, the SIP mechanism merges the information from these sources in order to optimize the precision of the controlled object's state estimate. It is demonstrated, based on scratching in decerebrate cats as an example of the spinal control of goal-directed movements, that the results of computer modeling agree with the experimental observations related to the SMCS's reactions to phasic and tonic peripheral afferent stimuli. It is also shown that the functional requirements imposed by the mathematical model of the SMCS comply with the current knowledge about the related properties of spinal neuronal circuitry. The crucial role of the spinal presynaptic inhibition mechanism in the neuronal implementation of SIP is elucidated. Important differences between the IM and a state predictor employed for compensating for a neural reflex time delay are discussed.

Statistics

Shimansky, Y.P. (2006) Continuous significant linear dimensionality: geometric interpretation and statistical characteristics. Computational Statistics & Data Analysis. 50(10): 2863-2877. [abstract]
A continuous measure of the degree of mutual linear independence between the vectors of a given dataset, continuous significant dimensionality (CSD),was introduced in a previous paper. In this report a simple and clear geometric interpretation of the CSD formula is determined, which has allowed to describe a parametric class of suchmeasures and present the conventional linear dimensionality as a limiting case of that class. Two additional methods for measuring CSD are developed and their properties are investigated bothth eoretically and based on computational experiments. A special attention is devoted to measuring CSD of datasets withdif ferent levels of noise. An efficient method for eliminating the impact of such noise on the CSD estimate is developed. It is demonstrated that the three methods for CSD measurement have different statistical characteristics and the scopes of their practical application are outlined.

Shimansky, Y.P. (2000) Continuous measure of significant linear dimensionality of a waveform set. Computational Statistics & Data Analysis. 35: 1-10. [abstract]
A novel method for measuring the degree of mutual linear independence between the waveforms of a given set is introduced as a natural result of combining principal component and correlation analyses. The proposed measure, continuous significant dimensionality (CSD), inherits its properties from both the dimensionality of a vector set as defined in linear algebra and the correlation between two waveforms. A simple formula for CSD based on the eigenvalues of the correlation matrix is suggested. It is shown that CSD is a continuous function of the cross-correlation coefficients, and it can be viewed as interpolating the dimensionality of a waveform set between cases in which the set contains an orthogonal basis. There are two important advantages of the CSD measure over the known integer estimate of significant linear dimensionality. First, the CSD formula does not depend on any subjective parameter such as "significance level". Second, the CSD allows the detection of relatively small differences in the degree of linear independence of waveforms between different sets. A complementary measure of the degree of linear interdependence between the waveforms of a given set also is described. The application of these measures is illustrated in the example of kinematic data processing.

Neuroscience

Dounskaia N., & Shimansky Y. (2020) Generalization of the resource-rationality principle to neural control of goal-directed movements. Behavioral and Brain Sciences, 43, E10. doi:10.1017/S0140525X19001559

Shimansky Y., Dounskaia N. (2018) Inclusion of neural effort in cost function can explain perceptual decision suboptimality. Behavioral and Brain Sciences, 41, E242.

Dounskaia N., Shimansky Y. (2016) Strategy of arm movement control is determined by minimization of neural effort for joint coordination. Exp Brain Res. 234(6):1335-1350.

Rand M.K., Shimansky Y.P. (2013) Two-phase strategy of neural control for planar reaching movements: II. Relation to spatiotemporal characteristics. Exp Brain Res. 230(1):1-13.

Shimansky Y.P., Rand M.K. (2013) Two-phase strategy of controlling motor coordination determined by task performance optimality. Biological Cybernetics. 107(1): 107-129. [abstract]
A quantitative model of optimal coordination between hand transport and grip aperture has been derived in our previous studies of reach-to-grasp movements without utilizing explicit knowledge of the optimality criterion or motor plant dynamics. The model’s utility for experimental data analysis has been demonstrated. Here we show how to generalize this model for a broad class of reaching-type, goal-directed movements. The model allows for measuring the variability of motor coordination and studying its dependence on movement phase. The experimentally found characteristics of that dependence imply that execution noise is low and does not affect motor coordination significantly. From those characteristics it is inferred that the cost of neural computations required for information acquisition and processing is included in the criterion of task performance optimality as a function of precision demand for state estimation and decision making. The precision demand is an additional optimized control variable that regulates the amount of neurocomputational resources activated dynamically. It is shown that an optimal control strategy in this case comprises two different phases. During the initial phase, the cost of neural computations is significantly reduced at the expense of reducing the demand for their precision, which results in speed-accuracy tradeoff violation and significant inter-trial variability of motor coordination. During the final phase, neural computations and thus motor coordination are considerably more precise to reduce the cost of errors in making a contact with the target object. The generality of the optimal coordination model and the two-phase control strategy is illustrated on several diverse examples.

Rand M.K., Shimansky Y.P. (2013) Two-phase strategy of neural control for planar reaching movements: I. XY coordination variability and its relation to end-point variability. Exp Brain Res. 225(1): 55-73.

Rand M.K., Shimansky Y.P., Hossain A.B.M.I., Stelmach G.E. (2010) Phase dependence of transport–aperture coordination variability reveals control strategy of reach-to-grasp movements. Experimental Brain Research. 207(1-2):49-63.

Rand M.K., Lemay M., Squire L.M., Shimansky Y.P., Stelmach G.E. (2009) Control of aperture closure initiation during reach-to-grasp movements under manipulations of visual feedback and trunk involvement in Parkinson’s disease. Experimental Brain Research. 201(3):509-525.

Rand M.K., Shimansky Y.P., Hossain A.B.M.I., and Stelmach G.E. (2008) Quantitative model of transport-aperture coordination during reach-to-grasp movements. Exp. Brain Res. 188: 263-274.

Rand M.K., Lemay M., Squire L.M., Shimansky Y.P., Stelmach G.E. (2007) Role of vision in aperture closures control during reach-to-grasp movements. Experimental Brain Research. 207: 49–63.

Goble J., Zhang Y., Shimansky Y., Sharma S., Dounskaia N. (2007) Directional biases reveal utilization of arm’s biomechanical properties for optimization of motor behavior. J. Neurophysiol. 98: 1240-1252.

Rand M.K., Smiley-Oyen A.L., Shimansky Y.P., Bloedel J.R. and Stelmach G.E. (2006) Control of aperture closure during reach-to-grasp movements in Parkinson’s disease. Exp. Brain Res. 168:131-42.

Rand, M.K., Shimansky, Y., Stelmach, G.E., Y., and Bloedel, J.R. (2004) Adaptation of reach-to-grasp movements in response to force perturbations. Exp. Brain Res. 154:50-65.

Shimansky, Y., Wang, J-J., Bauer, R.A., Bracha, V., and Bloedel, J.R. (2004) On-line compensation for perturbations of a reaching movement is cerebellar dependent. Exp. Brain Res. 155(2): 156-72. [abstract]
Although the cerebellum has been shown to be critical for the acquisition and retention of adaptive modifications in certain reflex behaviors, this structure’s role in the learning of motor skills required to execute complex voluntary goal-directed movements still is unclear. This study explores this issue by analyzing the effects of inactivating the interposed and dentate cerebellar nuclei on the adaptation required to compensate for an external elastic load applied during a reaching movement. We show that cats with these nuclei inactivated can adapt to predictable perturbations of the forelimb during a goaldirected reach by including a compensatory component in the motor plan prior to movement initiation. In contrast, when comparable compensatory modifications must be triggered on-line because the perturbations are applied in randomized trials (i.e., unpredictably), such adaptive responses cannot be executed or reacquired after the interposed and dentate nuclei are inactivated. These findings provide the first demonstration of the conditiondependent nature of the cerebellum’s contribution to the learning of a specific volitional task.

Rand, M.K., Shimansky, Y., Stelmach, G.E., Y., Bracha, V., and Bloedel, J.R. (2000) Effects of accuracy constraints on reach-to-grasp movements in cerebellar patients. Exp. Brain Res. 135: 179-188.

Shimansky, Y.P. (2000) Spinal motor control system incorporates an internal model of limb dynamics. Biological Cybernetics. 83: 379-389.

Shimansky, YP, Timmann D, Kolb FP, Diener HC, Bloedel JR. (1999) A role of the cerebellum in perceiving patterns in different sense modalities. In: Killeen PR, Uttal WR (Eds.) Fechner Day 99: The end of 20th Century Psychophysics. Proceedings of the 15-th annual meeting of the International Society for Psychophysics, Tempe, AZ, USA: The International Society of Psychophysics, pp 62-67.

Wang, J.-J., Shimansky Y, Bracha V, Bloedel JR. (1998) Effects of cerebellar nuclear inactivation on the learning of a complex forelimb movement in cats. J. Neurophysiol. 79: 2447-2459.

Shimansky, Y., Saling, M., Wunderlich, D., Bracha, V., Stelmach, G., and Bloedel, J.R. (1997) Impaired capacity of cerebellar patients to perceive and learn two-dimensional shapes based on kinesthetic cues. Learning and Memory. 4:36-48.

Milak, M., Shimansky, Yu., Bracha, V., and Bloedel, J.R. (1997) Effects of inactivating individual cerebellar nuclei on the performance and retention of a complex, operantly conditioned forelimb movement. J. Neurophysiol. 78:939-958.

Bloedel, J.R., Bracha, V., Milak, M., and Shimansky, Y. (1997) Cerebellar contributions to the acquisition and execution of learned reflex and volitional movements. In: De Zeeuw, C.I. (ed.) The Cerebellum: From Structure to Control. Elsevier: Amsterdam, pp. 499-509.

Bloedel, J.R., Bracha, V., Shimansky, Y., and Milak, M.S. (1996) The role of the cerebellum in the acquisition of complex, volitional forelimb movements. In: Bloedel, J.R., Ebner, T.J., and Wise, S.P. (eds.) Acquisition of Motor Behavior in Vertebrates. MIT Press: Boston, pp. 319-342.

Timmann, D., Shimansky, Yu., Larson, P.S., Wunderlich, D.A., Stelmach, G.E., and Bloedel, J.R. (1996) Visuomotor learning in cerebellar patients. Behav. Brain Res. 81:99-113.

Baev, K.V., Shimansky, Y.P. (1992) Principles of organization of neural systems controlling automatic movements in animals. Progress in Neurobiology. 39: 45-112.