Stability and encoding properties of two-layer nonlinear feedback neural networks are examined. Bidirectionality, forward and backard information flow, is introduced in neural nets to produce two-way associative search for stored associations (A, B, ). Passing information through M gives one direction; passing it through its transpose M r gives the other. A bidirectional associative memory. (BAM) behaves as a hetero- associative content addressable memory (CAM), storing and recalling the vector pairs (A1, Bi),-..,(Am Bin) , where .4 {0,1}"and B We prove that every n-by-p matrix M is a bidirectionally stable heteroas- sociative CAM for both binary/bipolar and continuous neurons a, and hi. When the BAM neurons are activated, the network quickly evolves to a stable state of two-pattern reverberation, or resonance. The stable reverberation corresponds to a system energy local minimum. Heteroassociafive inlormation is encoded iu a BAM by summing correlation matrices. The BAM storage capact .ty for reliable recall is roughly m < niin(n, p). No more heteroassociafive pairs can be 'reliably stored and recalled than the lesser of the dimensions of the pattern spaces (0,1 }"and 0,1 } P. The Appendix shos that it is better on average to use bipolar {- 1,i} coding than binary. {0,1 } coding of heteroassociative pairs (.4, B,). BAM encoding and decoding are combined in the adaptive BAM, which extends global bidirectional stabflit), to realtime unsupervised learning. Temporal patterns (AE,--., A,,) are represented as ordered lists of binary/bipolar vectors and stored in a temporal associative memory (TAM) n-by- matrix M as a limit cycle of the dynamical system. Forward recall proceeds through M, backward recall through M r . Temporal patterns are stored by summing contiguous bipolar...

The concepts of declarative memory and procedural memory have been used to distinguish two basic types of learning. A neural network model suggests how such memory processes work together as recognition learning, reinforcement learning, and sensory-motor learning take place during adaptive behaviors. To coordinate these processes, the hippocampal formation and cerebellum each contain circuits that learn to adaptively time their outputs. Within the model, hippocampal timing helps to maintain attention on motivationally salient goal objects during variable task-related delays, and cerebellar timing controls the release of conditioned responses. This property is part of the model's description of how cognitive-emotional interactions focus attention on motivationally valued cues, and how this process breaks down due to hippocampal ablation. The model suggests that the hippocampal mechanisms that help to rapidly draw attention to salient cues could prematurely release motor commands were no...

This article develops a neural model of how sharp disparity tuning can arise through experience-dependent development of cortical complex cells. This learning process clarifies how complex cells can binocularly match left and right eye image features with the same contrast polarity, yet also pool signals with opposite contrast polarities. Antagonistic rebounds between LGN ON and OFF cells and cortical simple cells sensitive to opposite contrast polarities enable anticorrelated simple cells to learn to activate a shared set of complex cells. Feedback from binocularly tuned cortical cells to monocular LGN cells is proposed to carry out a matching process that dynamically stabilizes the learning process. This feedback represents a type of matching process that is elaborated at higher visual processing areas into a volitionally controllable type of attention. We show stable learning when both of these properties hold. Learning adjusts the initially coarsely tuned disparity preference to ma...

How does the saccadic movement system select a target when visual, auditory, and planned movement commands differ? How do retinal, head-centered, and motor error coordinates interact during the selection process? Recent data on superior colliculus (SC) reveal a spreading wave of activation across buildup cells the peak activity of which covaries with the current gaze error. In contrast, the locus of peak activity remains constant at burst cells, whereas their activity level decays with residual gaze error. A neural model answers these questions and simulates burst and buildup responses in visual, overlap, memory, and gap tasks. The model also simulates data on multimodal enhancement and suppression of activity in the deeper SC layers and suggests a functional role for NMDA receptors in this region. In particular, the model suggests how auditory and planned saccadic target positions become aligned and compete

this article we introduce an unsupervisedneural architecture for the control of a mobile robot. The mobile robot to be controlled is organized in a tricycle structure. Movement is performed by selection of angular velocities for the motors attached to the two propulsive wheels, as shown in figure 1. Following an initial learning phase, the controller architecture allows movement between arbitrary points through exteroceptive or visual information. It is important to note that rather than learning explicit trajectories, the controller learns the relationship between angular velocities and the magnitude and direction of the resulting movement. This approach solves the inverse kinematic problem, so that visual information in spatial coordinates can generate the appropriate wheel angular velocities to move the mobile robot to a desired goal. The main characteristic of this architecture that distinguishes it from other neural controllers is that it does not require supervision during the training phase. Supervised neural network models [5, 12] require user knowledge to ensure that the environment during learning is statistically representative of the environment encountered during normal operation. This problem becomes critical when it is necessary to operate in unstructured environments or when the conditions of operation change. Another characteristic is that the system can learn continuously, i.e., it is not necessary to separate the learning phase from the operational phase. This property affords incremental and continuous learning, and adaptation to plant changes such as wear and tear of wheels, and other miscalibrations that may result from normal operation. In the next section we summarize the main characteristics of the Vector Associative Map (VAM) model, on which th...

Many of our sensory and motor organs have linearly ordered components, for example the fingers on a hand, the tonotopic organization of the auditory system, the successive joints on arms and legs, the spine, etc. This paper begins a discussion of some nonlinear networks which can learn complicated spatiotemporal patterns

This report constitutes an expanded version of a presentation given by the author at the 1993 European Control Conference (short course on "Neural Nets for Control"). The first part places neurocontrol techniques in a general learning control framework. The second part of the report, which is essentially independent of the first, briefly surveys several basic theoretical results regarding neural networks.

Visually guided navigation through a cluttered natural scene is a challenging problem that animals and humans accomplish with ease. The ViSTARS neural model proposes how primates use motion information to segment objects and determine heading for purposes of goal approach and obstacle avoidance in response to video inputs from real and virtual environments. The model produces trajectories similar to those of human navigators. It does so by predicting how computationally complementary processes in cortical areas MT- /MSTv and MT + /MSTd compute object motion for tracking and self-motion for navigation, respectively. The model retina responds to transients in the input stream. Model V1 generates a local speed and direction estimate. This local motion estimate is ambiguous due to the neural aperture problem. Model MT + interacts with MSTd via an attentive feedback loop to compute accurate heading estimates in MSTd that quantitatively simulate properties of human heading estimation data. Model MT- interacts with MSTv via an attentive feedback loop to compute accurate estimates of speed, direction and position of moving objects. This object information is combined with heading information to produce steering decisions wherein goals behave like attractors and obstacles behave like repellers. These steering decisions lead to navigational trajectories that closely match human performance.

Abstract not found

In order to properly care for critically ill patients in the intensive care unit (ICU), clinicians must be aware of hemodynamic patterns. In a typical ICU a variety of physiologic measurements are made continuously and intermittently in an attempt to provide clinicians with the most accurate and precise data needed for recognizing such patterns. However, the data are disjointed, yielding little information beyond that provided by instantaneous high/low limit checking. While instantaneous limit checking is useful for determining immediate dangers, it does not provide much information about temporal patterns. As a result, the clinician is left to manually sift through an excess of data in the interest of generating information. In this study, an arrangement of selforganizing artificial neural networks (ANNs) was used to automate the discovery, recognition, and prediction of hemodynamic patterns in real-time. It is shown that the network is capable of recognizing the same hemodynamic patt...