Recent interest in human-level intelligence suggests a rethink of the role of machine learning in computational intelligence. We argue that without cognitive learning the goal of achieving human-level synthetic intelligence is far from completion. Here we review the principles underlying human learning and memory, and identify three of them, i.e., continuity, glocality, and compositionality, as the most fundamental to human-level machine learning. We then propose the recently-developed hypernetwork model as a candidate architecture for cognitive learning and memory. Hypernetworks are a random hypergraph structure higher-order probabilistic relations of data by an evolutionary self-organizing process based on molecular selfassembly. The chemically-based massive interaction for information organization and processing in the molecular hypernetworks, referred to as hyperinteractionism, is contrasted with the symbolist, connectionist, and dynamicist approaches to mind and intelligence. We demonstrate the generative learning capability of the hypernetworks to simulate linguistic recall memory, visual imagery, and language-vision crossmodal translation based on a video corpus of movies and dramas in a multimodal memory game environment. We also offer prospects for the hyperinteractionistic molecular mind approach to a unified theory of cognitive learning.

Abstract — Creative thinking is one of the hallmarks of human-level competence. Although it is still a poorly understood subject speculative ideas about brain processes involved in creative thinking may be implemented in computational models. A review of different approaches to creativity, insight and intuition is presented. Two factors are essential for creativity: imagination and selection or filtering. Imagination should be constrained by experience, while filtering in the case of creative use of words may be based on semantic and phonological associations. Analysis of brain processes involved in invention of new words leads to practical algorithms that create many interesting and novel names associated with a set of keywords. I.

Abstract. Neurocognitive processes responsible for representation of meaning and understanding of words are investigated. First a review of current knowledge about word representation, recent experiments linking it to associative memory and to right hemisphere synchronous activity is presented. Various conjectures on how meaning arises and how reasoning and problem solving is done are presented. These inspirations are used to make systematic approximation to spreading activation in semantic memory networks. Using hierarchical ontologies representations of short texts are enhanced and it is shown that highdimensional vector models may be treated as a snapshot approximation of the neural activity. Clustering short medical texts into different categories is greatly enhanced by this process, thus facilitating understanding of the text. 1

Abstract — Neural networks have been proposed as an ideal cognitive modeling methodology to deal with the symbol grounding problem. More recently, such neural network approaches have been incorporated in studies based on cognitive agents and robots. In this paper we present a new model of symbol grounding transfer in cognitive robots. Language learning simulations demonstrate that robots are able to acquire new action concepts via linguistic instructions. This is achieved by autonomously transferring the grounding from directly grounded action names to new higher-order composite actions. The robot’s neural network controller permits such a grounding transfer. The implications for such a modeling approach in cognitive science and autonomous robotics are discussed.

We extend the in-situ Hebbian-linked SOMs network by Miikkulainen to come up with two unsupervised neural networks that learn the mapping between the individual modes of a multimodal dataset. The first network, the single-pass Hebbian linked SOMs network, extends the in-situ Hebbian-linked SOMs network by enabling the Hebbian link weights to be computed through one-shot learning. The second network, a modified counterpropagation network, extends the unsupervised learning of crossmodal mappings by making it possible for only one self-organising map to implement the crossmodal mapping. The two proposed networks each have a smaller computation time and achieve lower crossmodal mean squared errors than the in-situ Hebbian-linked SOMs network when assessed on two bimodal datasets, an audio-acoustic speech utterance dataset and a phonological-semantics child utterance dataset. Of the three network architectures, the modified counterpropagation network achieves the highest percentage of correct classifications comparable to that of the LVQ-2 algorithm by Kohonen and the neural network for category learning by de Sa and Ballard in classification tasks using the audio-acoustic speech utterance dataset.

Neural Modeling Field Theory is based on the principle of associating lowerlevel signals (e.g., inputs, bottom-up signals) with higher-level concept-models (e.g. internal representations, categories/concepts, top-down signals) avoiding the combinatorial complexity inherent to such a task. In this paper we present an extension of the Modeling Field Theory neural network for the classification of objects. Simulations show that (i) the system is able to dynamically adapt when an additional feature is introduced during learning, (ii) that this algorithm can be applied to the classification of action patterns in the context of cognitive robotics and (iii) that it is able to classify multi-feature objects from complex stimulus set. The use of Modeling Field Theory for studying the integration of language and cognition in robots is discussed.

Abstract—Putative brain processes responsible for understanding language are based on spreading activation in semantic networks, providing enhanced representations that involve concepts not found directly in the text. Approximation of this process is of great practical and theoretical interest. Vector model should reflect activations of various concepts in the brain spreading through associative network. Medical ontologies are used to select concepts of specific semantic type and add to each of them related concepts, providing expanded vector representations. The process is constrained by selection of useful extensions for the classification task. Short hospital discharge summaries are used to illustrate how this process works on a real, very noisy data. Results show significantly improved clustering and classification accuracy. A practical approach to mapping of associative networks of the brain involved in processing of specific concepts is presented. I.

www.his.sunderland.ac.uk By using neurocognitive evidence on mirror neuron system concepts the MirrorBot project has developed neural models for intelligent robot behaviour. These models employ diverse learning approaches such as reinforcement learning, self-organisation and associative learning to perform cognitive robotic operations such as language grounding in actions, object recognition, localisation and docking. In this paper we describe architectures based on an associative self-organising framework which were designed to combine multimodal inputs of language, vision and motor programs to produce complex robot behaviours. 1.

Learning by demonstration and learning by instruction offers a potentially more powerful paradigm than programming robots directly for specific tasks. Learning in humans or primates substantially benefits from demonstration of actions or instruction by language in the appropriate context and there is initial neurocognitive cortical evidence for such processes. Cortical assemblies have been identified in the cortex that activate in response to the performance of motor tasks at a semantic level. This evidence supports that such mirror neuron assemblies are involved in actions, observing actions and communicating actions. Furthermore, neurocognitive evidence supports that cell assemblies are activated in different regions of the brain dependent on the action type being processed. Based on this neurocognitive evidence we have begun to design a neural robot in the MirrorBot project that is based on multimodal integration and topological organisation of actions using associative memory. As part of these studies in this paper we describe a self-organising model that clusters actions into different locations dependent on the body part they are associated with. In particular, we use actual sensor readings from the MIRA robot to represent semantic features of the action verbs. Furthermore, ongoing work focuses on integration of motor, vision and language representations for learning from demonstration and language instruction.

Abstract. For this special session of EU projects in the area of NeuroIT, we will review the progress of the MirrorBot project with special emphasis on its relation to reinforcement learning and future perspectives. Models inspired by mirror neurons in the cortex, while enabling a system to understand its actions, also help in the solving of the curse of dimensionality problem of reinforcement learning. Reinforcement learning, which is primarily linked to the basal ganglia, is a powerful method to teach an agent such as a robot a goal-directed action strategy. Its limitation is mainly that the perceived situation has to be mapped to a state space, which grows exponentially with input dimensionality. Cortex-inspired computation can alleviate this problem by pre-processing sensory information and supplying motor primitives that can act as modules for a superordinate reinforcement learning scheme. 1