Spatio-temporal credit assignment in population learning

We present a model for plasticity induction in reinforcement learning which is based on a cascade of synaptic memory traces. In the cascade of these so called eligibility traces presynaptic input is first corre lated with postsynaptic events, next with the behavioral decisions and finally with the external reinforcement. A population of leaky integrate and fire neurons endowed with this plasticity scheme is studied by simulation on different tasks. For operant co nditioning with delayed reinforcement, learning succeeds even when the delay is so large that the delivered reward reflects the appropriateness, not of the immediately preceeding response, but of a decision made earlier on in the stimulus - decision sequence . So the proposed model does not rely on the temporal contiguity between decision and pertinent reward and thus provides a viable means of addressing the temporal credit assignment problem. In the same task, learning speeds up with increasing population si ze, showing that the plasticity cascade simultaneously addresses the spatial problem of assigning credit to the different population neurons. Simulations on other task such as sequential decision making serve to highlight the robustness of the proposed sch eme and, further, contrast its performance to that of temporal difference based approaches to reinforcement learning.

Spatio-Temporal Credit Assignment in Neuronal Population Learning

n learning from trial and error, animals need to relate behavioral decisions to environmental reinforcement even though it may be difficult to assign credit to a particular decision when outcomes are uncertain or subject to delays. When considering the biophysical basis of learning, the credit-assignment problem is compounded because the behavioral decisions themselves result from the spatio-temporal aggregation of many synaptic releases. We present a model of plasticity induction for reinforcement learning in a population of leaky integrate and fire neurons which is based on a cascade of synaptic memory traces. Each synaptic cascade correlates presynaptic input first with postsynaptic events, next with the behavioral decisions and finally with external reinforcement. For operant conditioning, learning succeeds even when reinforcement is delivered with a delay so large that temporal contiguity between decision and pertinent reward is lost due to intervening decisions which are themselves subject to delayed reinforcement. This shows that the model provides a viable mechanism for temporal credit assignment. Further, learning speeds up with increasing population size, so the plasticity cascade simultaneously addresses the spatial problem of assigning credit to synapses in different population neurons. Simulations on other tasks, such as sequential decision making, serve to contrast the performance of the proposed scheme to that of temporal difference-based learning. We argue that, due to their comparative robustness, synaptic plasticity cascades are attractive basic models of reinforcement learning in the brain.

Dopamine-related deficit in reward learning after catecholamine depletion in unmedicated, remitted subjects with bulimia nervosa

Disturbances in reward processing have been implicated in bulimia nervosa (BN). Abnormalities in processing reward-related stimuli might be linked to dysfunctions of the catecholaminergic neurotransmitter system, but findings have been inconclusive. A powerful way to investigate the relationship between catecholaminergic function and behavior is to examine behavioral changes in response to experimental catecholamine depletion (CD). The purpose of this study was to uncover putative catecholaminergic dysfunction in remitted subjects with BN who performed a reinforcement-learning task after CD. CD was achieved by oral alpha-methyl-para-tyrosine (AMPT) in 19 unmedicated female subjects with remitted BN (rBN) and 28 demographically matched healthy female controls (HC). Sham depletion administered identical capsules containing diphenhydramine. The study design consisted of a randomized, double-blind, placebo-controlled crossover, single-site experimental trial. The main outcome measures were reward learning in a probabilistic reward task analyzed using signal-detection theory. Secondary outcome measures included self-report assessments, including the Eating Disorder Examination-Questionnaire. Relative to healthy controls, rBN subjects were characterized by blunted reward learning in the AMPT-but not in placebo-condition. Highlighting the specificity of these findings, groups did not differ in their ability to perceptually distinguish between stimuli. Increased CD-induced anhedonic (but not eating disorder) symptoms were associated with a reduced response bias toward a more frequently rewarded stimulus. In conclusion, under CD, rBN subjects showed reduced reward learning compared with healthy control subjects. These deficits uncover disturbance of the central reward processing systems in rBN related to altered brain catecholamine levels, which might reflect a trait-like deficit increasing vulnerability to BN.

Spike-based decision learning of nash equilibria in two-player games

Humans and animals face decision tasks in an uncertain multi-agent environment where an agent's strategy may change in time due to the co-adaptation of others strategies. The neuronal substrate and the computational algorithms underlying such adaptive decision making, however, is largely unknown. We propose a population coding model of spiking neurons with a policy gradient procedure that successfully acquires optimal strategies for classical game-theoretical tasks. The suggested population reinforcement learning reproduces data from human behavioral experiments for the blackjack and the inspector game. It performs optimally according to a pure (deterministic) and mixed (stochastic) Nash equilibrium, respectively. In contrast, temporal-difference(TD)-learning, covariance-learning, and basic reinforcement learning fail to perform optimally for the stochastic strategy. Spike-based population reinforcement learning, shown to follow the stochastic reward gradient, is therefore a viable candidate to explain automated decision learning of a Nash equilibrium in two-player games.

Learning flexible sensori-motor mappings in a complex network

Given the complex structure of the brain, how can synaptic plasticity explain the learning and forgetting of associations when these are continuously changing? We address this question by studying different reinforcement learning rules in a multilayer network in order to reproduce monkey behavior in a visuomotor association task. Our model can only reproduce the learning performance of the monkey if the synaptic modifications depend on the pre- and postsynaptic activity, and if the intrinsic level of stochasticity is low. This favored learning rule is based on reward modulated Hebbian synaptic plasticity and shows the interesting feature that the learning performance does not substantially degrade when adding layers to the network, even for a complex problem.

Code-specific learning rules improve action selection by populations of spiking neurons

Population coding is widely regarded as a key mechanism for achieving reliable behavioral decisions. We previously introduced reinforcement learning for population-based decision making by spiking neurons. Here we generalize population reinforcement learning to spike-based plasticity rules that take account of the postsynaptic neural code. We consider spike/no-spike, spike count and spike latency codes. The multi-valued and continuous-valued features in the postsynaptic code allow for a generalization of binary decision making to multi-valued decision making and continuous-valued action selection. We show that code-specific learning rules speed up learning both for the discrete classification and the continuous regression tasks. The suggested learning rules also speed up with increasing population size as opposed to standard reinforcement learning rules. Continuous action selection is further shown to explain realistic learning speeds in the Morris water maze. Finally, we introduce the concept of action perturbation as opposed to the classical weight- or node-perturbation as an exploration mechanism underlying reinforcement learning. Exploration in the action space greatly increases the speed of learning as compared to exploration in the neuron or weight space.

Learning by the Dendritic Prediction of Somatic Spiking

Recent modeling of spike-timing-dependent plasticity indicates that plasticity involves as a third factor a local dendritic potential, besides pre- and postsynaptic firing times. We present a simple compartmental neuron model together with a non-Hebbian, biologically plausible learning rule for dendritic synapses where plasticity is modulated by these three factors. In functional terms, the rule seeks to minimize discrepancies between somatic firings and a local dendritic potential. Such prediction errors can arise in our model from stochastic fluctuations as well as from synaptic input, which directly targets the soma. Depending on the nature of this direct input, our plasticity rule subserves supervised or unsupervised learning. When a reward signal modulates the learning rate, reinforcement learning results. Hence a single plasticity rule supports diverse learning paradigms.

Exact Combinatorial Optimization with Graph Convolutional Neural Networks

Combinatorial optimization problems are typically tackled by the branch-and-bound paradigm. We propose to learn a variable selection policy for branch-and-bound in mixed-integer linear programming, by imitation learning on a diversified variant of the strong branching expert rule. We encode states as bipartite graphs and parameterize the policy as a graph convolutional neural network. Experiments on a series of synthetic problems demonstrate that our approach produces policies that can improve upon expert-designed branching rules on large problems, and generalize to instances significantly larger than seen during training.

Integrated communication and learning towards autonomous mobile radio networks

The integration of distributed and ubiquitous intelligence has emerged over the last years as the mainspring of transformative advancements in mobile radio networks. As we approach the era of “mobile for intelligence”, next-generation wireless networks are poised to undergo significant and profound changes. Notably, the overarching challenge that lies ahead is the development and implementation of integrated communication and learning mechanisms that will enable the realization of autonomous mobile radio networks. The ultimate pursuit of eliminating human-in-the-loop constitutes an ambitious challenge, necessitating a meticulous delineation of the fundamental characteristics that artificial intelligence (AI) should possess to effectively achieve this objective. This challenge represents a paradigm shift in the design, deployment, and operation of wireless networks, where conventional, static configurations give way to dynamic, adaptive, and AI-native systems capable of self-optimization, self-sustainment, and learning. This thesis aims to provide a comprehensive exploration of the fundamental principles and practical approaches required to create autonomous mobile radio networks that seamlessly integrate communication and learning components. The first chapter of this thesis introduces the notion of Predictive Quality of Service (PQoS) and adaptive optimization and expands upon the challenge to achieve adaptable, reliable, and robust network performance in dynamic and ever-changing environments. The subsequent chapter delves into the revolutionary role of generative AI in shaping next-generation autonomous networks. This chapter emphasizes achieving trustworthy uncertainty-aware generation processes with the use of approximate Bayesian methods and aims to show how generative AI can improve generalization while reducing data communication costs. Finally, the thesis embarks on the topic of distributed learning over wireless networks. Distributed learning and its declinations, including multi-agent reinforcement learning systems and federated learning, have the potential to meet the scalability demands of modern data-driven applications, enabling efficient and collaborative model training across dynamic scenarios while ensuring data privacy and reducing communication overhead.

Multiagent system for adaptive strategy formulation in electricity markets

Electricity markets are complex environments with very particular characteristics. MASCEM is a market simulator developed to allow deep studies of the interactions between the players that take part in the electricity market negotiations. This paper presents a new proposal for the definition of MASCEM players’ strategies to negotiate in the market. The proposed methodology is multiagent based, using reinforcement learning algorithms to provide players with the capabilities to perceive the changes in the environment, while adapting their bids formulation according to their needs, using a set of different techniques that are at their disposal.

Bid definition method for electricity markets based on an adaptive multiagent system

Electricity markets are complex environments with very particular characteristics. MASCEM is a market simulator developed to allow deep studies of the interactions between the players that take part in the electricity market negotiations. This paper presents a new proposal for the definition of MASCEM players’ strategies to negotiate in the market. The proposed methodology is multiagent based, using reinforcement learning algorithms to provide players with the capabilities to perceive the changes in the environment, while adapting their bids formulation according to their needs, using a set of different techniques that are at their disposal. Each agent has the knowledge about a different method for defining a strategy for playing in the market, the main agent chooses the best among all those, and provides it to the market player that requests, to be used in the market. This paper also presents a methodology to manage the efficiency/effectiveness balance of this method, to guarantee that the degradation of the simulator processing times takes the correct measure.

Distributed conditional computation

L'objectif de cette thèse est de présenter différentes applications du programme de recherche de calcul conditionnel distribué. On espère que ces applications, ainsi que la théorie présentée ici, mènera à une solution générale du problème d'intelligence artificielle, en particulier en ce qui a trait à la nécessité d'efficience. La vision du calcul conditionnel distribué consiste à accélérer l'évaluation et l'entraînement de modèles profonds, ce qui est très différent de l'objectif usuel d'améliorer sa capacité de généralisation et d'optimisation. Le travail présenté ici a des liens étroits avec les modèles de type mélange d'experts. Dans le chapitre 2, nous présentons un nouvel algorithme d'apprentissage profond qui utilise une forme simple d'apprentissage par renforcement sur un modèle d'arbre de décisions à base de réseau de neurones. Nous démontrons la nécessité d'une contrainte d'équilibre pour maintenir la distribution d'exemples aux experts uniforme et empêcher les monopoles. Pour rendre le calcul efficient, l'entrainement et l'évaluation sont contraints à être éparse en utilisant un routeur échantillonnant des experts d'une distribution multinomiale étant donné un exemple. Dans le chapitre 3, nous présentons un nouveau modèle profond constitué d'une représentation éparse divisée en segments d'experts. Un modèle de langue à base de réseau de neurones est construit à partir des transformations éparses entre ces segments. L'opération éparse par bloc est implémentée pour utilisation sur des cartes graphiques. Sa vitesse est comparée à deux opérations denses du même calibre pour démontrer le gain réel de calcul qui peut être obtenu. Un modèle profond utilisant des opérations éparses contrôlées par un routeur distinct des experts est entraîné sur un ensemble de données d'un milliard de mots. Un nouvel algorithme de partitionnement de données est appliqué sur un ensemble de mots pour hiérarchiser la couche de sortie d'un modèle de langage, la rendant ainsi beaucoup plus efficiente. Le travail présenté dans cette thèse est au centre de la vision de calcul conditionnel distribué émis par Yoshua Bengio. Elle tente d'appliquer la recherche dans le domaine des mélanges d'experts aux modèles profonds pour améliorer leur vitesse ainsi que leur capacité d'optimisation. Nous croyons que la théorie et les expériences de cette thèse sont une étape importante sur la voie du calcul conditionnel distribué car elle cadre bien le problème, surtout en ce qui concerne la compétitivité des systèmes d'experts.

SmartForge: An event-driven architecture for simulator-based process optimization

In the metal industry, and more specifically in the forging one, scrap material is a crucial issue and reducing it would be an important goal to reach. Not only would this help the companies to be more environmentally friendly and more sustainable, but it also would reduce the use of energy and lower costs. At the same time, the techniques for Industry 4.0 and the advancements in Artificial Intelligence (AI), especially in the field of Deep Reinforcement Learning (DRL), may have an important role in helping to achieve this objective. This document presents the thesis work, a contribution to the SmartForge project, that was performed during a semester abroad at Karlstad University (Sweden). This project aims at solving the aforementioned problem with a business case of the company Bharat Forge Kilsta, located in Karlskoga (Sweden). The thesis work includes the design and later development of an event-driven architecture with microservices, to support the processing of data coming from sensors set up in the company's industrial plant, and eventually the implementation of an algorithm with DRL techniques to control the electrical power to use in it.

Híper-heurísticas com aprendizagem

A otimização nos sistemas de suporte à decisão atuais assume um carácter fortemente interdisciplinar relacionando-se com a necessidade de integração de diferentes técnicas e paradigmas na resolução de problemas reais complexos, sendo que a computação de soluções ótimas em muitos destes problemas é intratável. Os métodos de pesquisa heurística são conhecidos por permitir obter bons resultados num intervalo temporal aceitável. Muitas vezes, necessitam que a parametrização seja ajustada de forma a permitir obter bons resultados. Neste sentido, as estratégias de aprendizagem podem incrementar o desempenho de um sistema, dotando-o com a capacidade de aprendizagem, por exemplo, qual a técnica de otimização mais adequada para a resolução de uma classe particular de problemas, ou qual a parametrização mais adequada de um dado algoritmo num determinado cenário. Alguns dos métodos de otimização mais usados para a resolução de problemas do mundo real resultaram da adaptação de ideias de várias áreas de investigação, principalmente com inspiração na natureza - Meta-heurísticas. O processo de seleção de uma Meta-heurística para a resolução de um dado problema é em si um problema de otimização. As Híper-heurísticas surgem neste contexto como metodologias eficientes para selecionar ou gerar heurísticas (ou Meta-heurísticas) na resolução de problemas de otimização NP-difícil. Nesta dissertação pretende-se dar uma contribuição para o problema de seleção de Metaheurísticas respetiva parametrização. Neste sentido é descrita a especificação de uma Híperheurística para a seleção de técnicas baseadas na natureza, na resolução do problema de escalonamento de tarefas em sistemas de fabrico, com base em experiência anterior. O módulo de Híper-heurística desenvolvido utiliza um algoritmo de aprendizagem por reforço (QLearning), que permite dotar o sistema da capacidade de seleção automática da Metaheurística a usar no processo de otimização, assim como a respetiva parametrização. Finalmente, procede-se à realização de testes computacionais para avaliar a influência da Híper- Heurística no desempenho do sistema de escalonamento AutoDynAgents. Como conclusão genérica, é possível afirmar que, dos resultados obtidos é possível concluir existir vantagem significativa no desempenho do sistema quando introduzida a Híper-heurística baseada em QLearning.

Modèle informatique du coapprentissage des ganglions de la base et du cortex : l'apprentissage par renforcement et le développement de représentations

Tout au long de la vie, le cerveau développe des représentations de son environnement permettant à l’individu d’en tirer meilleur profit. Comment ces représentations se développent-elles pendant la quête de récompenses demeure un mystère. Il est raisonnable de penser que le cortex est le siège de ces représentations et que les ganglions de la base jouent un rôle important dans la maximisation des récompenses. En particulier, les neurones dopaminergiques semblent coder un signal d’erreur de prédiction de récompense. Cette thèse étudie le problème en construisant, à l’aide de l’apprentissage machine, un modèle informatique intégrant de nombreuses évidences neurologiques. Après une introduction au cadre mathématique et à quelques algorithmes de l’apprentissage machine, un survol de l’apprentissage en psychologie et en neuroscience et une revue des modèles de l’apprentissage dans les ganglions de la base, la thèse comporte trois articles. Le premier montre qu’il est possible d’apprendre à maximiser ses récompenses tout en développant de meilleures représentations des entrées. Le second article porte sur l'important problème toujours non résolu de la représentation du temps. Il démontre qu’une représentation du temps peut être acquise automatiquement dans un réseau de neurones artificiels faisant office de mémoire de travail. La représentation développée par le modèle ressemble beaucoup à l’activité de neurones corticaux dans des tâches similaires. De plus, le modèle montre que l’utilisation du signal d’erreur de récompense peut accélérer la construction de ces représentations temporelles. Finalement, il montre qu’une telle représentation acquise automatiquement dans le cortex peut fournir l’information nécessaire aux ganglions de la base pour expliquer le signal dopaminergique. Enfin, le troisième article évalue le pouvoir explicatif et prédictif du modèle sur différentes situations comme la présence ou l’absence d’un stimulus (conditionnement classique ou de trace) pendant l’attente de la récompense. En plus de faire des prédictions très intéressantes en lien avec la littérature sur les intervalles de temps, l’article révèle certaines lacunes du modèle qui devront être améliorées. Bref, cette thèse étend les modèles actuels de l’apprentissage des ganglions de la base et du système dopaminergique au développement concurrent de représentations temporelles dans le cortex et aux interactions de ces deux structures.