Showing 1–6 of 6 results for author: Pitti, A

On the Relationship Between Variational Inference and Auto-Associative Memory

Authors: Louis Annabi, Alexandre Pitti, Mathias Quoy

Abstract: In this article, we propose a variational inference formulation of auto-associative memories, allowing us to combine perceptual inference and memory retrieval into the same mathematical framework. In this formulation, the prior probability distribution onto latent representations is made memory dependent, thus pulling the inference process towards previously stored representations. We then study h… ▽ More In this article, we propose a variational inference formulation of auto-associative memories, allowing us to combine perceptual inference and memory retrieval into the same mathematical framework. In this formulation, the prior probability distribution onto latent representations is made memory dependent, thus pulling the inference process towards previously stored representations. We then study how different neural network approaches to variational inference can be applied in this framework. We compare methods relying on amortized inference such as Variational Auto Encoders and methods relying on iterative inference such as Predictive Coding and suggest combining both approaches to design new auto-associative memory models. We evaluate the obtained algorithms on the CIFAR10 and CLEVR image datasets and compare them with other associative memory models such as Hopfield Networks, End-to-End Memory Networks and Neural Turing Machines. △ Less

Submitted 14 October, 2022; originally announced October 2022.

A Predictive Coding Account for Chaotic Itinerancy

Authors: Louis Annabi, Alexandre Pitti, Mathias Quoy

Abstract: As a phenomenon in dynamical systems allowing autonomous switching between stable behaviors, chaotic itinerancy has gained interest in neurorobotics research. In this study, we draw a connection between this phenomenon and the predictive coding theory by showing how a recurrent neural network implementing predictive coding can generate neural trajectories similar to chaotic itinerancy in the prese… ▽ More As a phenomenon in dynamical systems allowing autonomous switching between stable behaviors, chaotic itinerancy has gained interest in neurorobotics research. In this study, we draw a connection between this phenomenon and the predictive coding theory by showing how a recurrent neural network implementing predictive coding can generate neural trajectories similar to chaotic itinerancy in the presence of input noise. We propose two scenarios generating random and past-independent attractor switching trajectories using our model. △ Less

Submitted 16 June, 2021; originally announced June 2021.

Bidirectional Interaction between Visual and Motor Generative Models using Predictive Coding and Active Inference

Authors: Louis Annabi, Alexandre Pitti, Mathias Quoy

Abstract: In this work, we build upon the Active Inference (AIF) and Predictive Coding (PC) frameworks to propose a neural architecture comprising a generative model for sensory prediction, and a distinct generative model for motor trajectories. We highlight how sequences of sensory predictions can act as rails guiding learning, control and online adaptation of motor trajectories. We furthermore inquire the… ▽ More In this work, we build upon the Active Inference (AIF) and Predictive Coding (PC) frameworks to propose a neural architecture comprising a generative model for sensory prediction, and a distinct generative model for motor trajectories. We highlight how sequences of sensory predictions can act as rails guiding learning, control and online adaptation of motor trajectories. We furthermore inquire the effects of bidirectional interactions between the motor and the visual modules. The architecture is tested on the control of a simulated robotic arm learning to reproduce handwritten letters. △ Less

Submitted 19 April, 2021; originally announced April 2021.

Towards a Neural Model for Serial Order in Frontal Cortex: a Brain Theory from Memory Development to Higher-Level Cognition

Authors: Alexandre Pitti, Mathias Quoy, Catherine Lavandier, Sofiane Boucenna, Wassim Swaileh, Claudio Weidmann

Abstract: In order to keep trace of information and grow up, the infant brain has to resolve the problem about where old information is located and how to index new ones. We propose that the immature prefrontal cortex (PFC) use its primary functionality of detecting hierarchical patterns in temporal signals as a second purpose to organize the spatial ordering of the cortical networks in the develo** brain… ▽ More In order to keep trace of information and grow up, the infant brain has to resolve the problem about where old information is located and how to index new ones. We propose that the immature prefrontal cortex (PFC) use its primary functionality of detecting hierarchical patterns in temporal signals as a second purpose to organize the spatial ordering of the cortical networks in the develo** brain itself. Our hypothesis is that the PFC detects the hierarchical structure in temporal sequences in the form of ordinal patterns and use them to index information hierarchically in different parts of the brain. Henceforth, we propose that this mechanism for detecting patterns participates in the ordinal organization development of the brain itself; i.e., the bootstrap** of the connectome. By doing so, it gives the tools to the language-ready brain for manipulating abstract knowledge and planning temporally ordered information; i.e., the emergence of symbolic thinking and language. We will review neural models that can support such mechanisms and propose new ones. We will confront then our ideas with evidence from developmental, behavioral and brain results and make some hypotheses, for instance, on the construction of the mirror neuron system, on embodied cognition, and on the capacity of learning-to-learn. △ Less

Submitted 30 December, 2021; v1 submitted 22 May, 2020; originally announced May 2020.

Autonomous learning and chaining of motor primitives using the Free Energy Principle

Authors: Louis Annabi, Alexandre Pitti, Mathias Quoy

Abstract: In this article, we apply the Free-Energy Principle to the question of motor primitives learning. An echo-state network is used to generate motor trajectories. We combine this network with a perception module and a controller that can influence its dynamics. This new compound network permits the autonomous learning of a repertoire of motor trajectories. To evaluate the repertoires built with our m… ▽ More In this article, we apply the Free-Energy Principle to the question of motor primitives learning. An echo-state network is used to generate motor trajectories. We combine this network with a perception module and a controller that can influence its dynamics. This new compound network permits the autonomous learning of a repertoire of motor trajectories. To evaluate the repertoires built with our method, we exploit them in a handwriting task where primitives are chained to produce long-range sequences. △ Less

Submitted 11 May, 2020; originally announced May 2020.

Ideas from Developmental Robotics and Embodied AI on the Questions of Ethics in Robots

Authors: Alexandre Pitti

Abstract: Advances in Artificial Intelligence and robotics are currently questioning theethical framework of their applications to deal with potential drifts, as well as the way inwhich these algorithms learn because they will have a strong impact on the behavior ofrobots and the type of robots. interactions with people. We would like to highlight someprinciples and ideas from cognitive neuroscience and dev… ▽ More Advances in Artificial Intelligence and robotics are currently questioning theethical framework of their applications to deal with potential drifts, as well as the way inwhich these algorithms learn because they will have a strong impact on the behavior ofrobots and the type of robots. interactions with people. We would like to highlight someprinciples and ideas from cognitive neuroscience and development sciences based on theimportance of the body for intelligence, contrary to the theory of the all-brain or all-algorithm, to represent the world and interacting with others, and their current applicationsin embodied AI and developmental robotics to propose models of architectures andmechanisms for agency, representation of the body, recognition of the intention of others,predictive coding, active inference, the role of feedback and error, imitation, artificialcuriosity and contextual learning. We will explain how these are important for the design ofautonomous systems and beyond what they can tell us for the ethics of systems. △ Less

Submitted 20 March, 2018; originally announced March 2018.