arXiv subjects

Ryota Kanai

Publications and source records attributed to Ryota Kanai.

35 records · Page 2Linked to original sources

arXiv2021

Experimental Evidence that Empowerment May Drive Exploration in Sparse-Reward Environments

Reinforcement Learning (RL) is known to be often unsuccessful in environments with sparse extrinsic rewards. A possible countermeasure is to endow RL agents with an intrinsic reward function, or 'intrinsic motivation', which rewards the agent based on certain features of the current sensor state. An intrinsic reward function based on the principle of empowerment assigns rewards proportional to the amount of control the agent has over its own sensors. We implemented a variation on a recently proposed intrinsically motivated agent, which we refer to as the 'curious' agent, and an empowerment-inspired agent. The former leverages sensor state encoding with a variational autoencoder, while the latter predicts the next sensor state via a variational information bottleneck. We compared the performance of both agents to that of an advantage actor-critic baseline in four sparse reward grid worlds. Both the empowerment agent and its curious competitor seem to benefit to similar extents from their intrinsic rewards. This provides some experimental support to the conjecture that empowerment can be used to drive exploration.

cs.AI↗

arXiv2021

A Technical Critique of Some Parts of the Free Energy Principle

We summarize the original formulation of the free energy principle, and highlight some technical issues. We discuss how these issues affect related results involving generalised coordinates and, where appropriate, mention consequences for and reveal, up to now unacknowledged, differences to newer formulations of the free energy principle. In particular, we reveal that various definitions of the "Markov blanket" proposed in different works are not equivalent. We show that crucial steps in the free energy argument which involve rewriting the equations of motion of systems with Markov blankets, are not generally correct without additional (previously unstated) assumptions. We prove by counterexample that the original free energy lemma, when taken at face value, is wrong. We show further that this free energy lemma, when it does hold, implies equality of variational density and ergodic conditional density. The interpretation in terms of Bayesian inference hinges on this point, and we hence conclude that it is not sufficiently justified. Additionally, we highlight that the variational densities presented in newer formulations of the free energy principle and lemma are parameterised by different variables than in older works, leading to a substantially different interpretation of the theory. Note that we only highlight some specific problems in the discussed publications. These problems do not rule out conclusively that the general ideas behind the free energy principle are worth pursuing.

q-bio.NC↗

arXiv2021

Deep Learning and the Global Workspace Theory

Recent advances in deep learning have allowed Artificial Intelligence (AI) to reach near human-level performance in many sensory, perceptual, linguistic or cognitive tasks. There is a growing need, however, for novel, brain-inspired cognitive architectures. The Global Workspace theory refers to a large-scale system integrating and distributing information among networks of specialized modules to create higher-level forms of cognition and awareness. We argue that the time is ripe to consider explicit implementations of this theory using deep learning techniques. We propose a roadmap based on unsupervised neural translation between multiple latent spaces (neural networks trained for distinct tasks, on distinct sensory inputs and/or modalities) to create a unique, amodal global latent workspace (GLW). Potential functional advantages of GLW are reviewed, along with neuroscientific implications.

cs.AI↗

arXiv2020

Non-trivial informational closure of a Bayesian hyperparameter

We investigate the non-trivial informational closure (NTIC) of a Bayesian hyperparameter inferring the underlying distribution of an identically and independently distributed finite random variable. For this we embed both the Bayesian hyper-parameter updating process and the random data process into a Markov chain. The original publication by Bertschinger et al. (2006) mentioned that NTIC may be able to capture an abstract notion of modeling that is agnostic to the specific internal structure of and existence of explicit representations within the modeling process. The Bayesian hyperparameter is of interest since it has a well defined interpretation as a model of the data process and at the same time its dynamics can be specified without reference to this interpretation. On the one hand we show explicitly that the NTIC of the hyperparameter increases indefinitely over time. On the other hand we attempt to establish a connection between a quantity that is a feature of the interpretation of the hyperparameter as a model, namely the information gain, and the one-step pointwise NTIC which is a quantity that does not depend on this interpretation. We find that in general we cannot use the one-step pointwise NTIC as an indicator for information gain. We hope this exploratory work can lead to further rigorous studies of the relation between NTIC and modeling.

cs.LG↗

arXiv2020

Information Closure Theory of Consciousness

Information processing in neural systems can be described and analysed at multiple spatiotemporal scales. Generally, information at lower levels is more fine-grained and can be coarse-grained in higher levels. However, information processed only at specific levels seems to be available for conscious awareness. We do not have direct experience of information available at the level of individual neurons, which is noisy and highly stochastic. Neither do we have experience of more macro-level interactions such as interpersonal communications. Neurophysiological evidence suggests that conscious experiences co-vary with information encoded in coarse-grained neural states such as the firing pattern of a population of neurons. In this article, we introduce a new informational theory of consciousness: Information Closure Theory of Consciousness (ICT). We hypothesise that conscious processes are processes which form non-trivial informational closure (NTIC) with respect to the environment at certain coarse-grained levels. This hypothesis implies that conscious experience is confined due to informational closure from conscious processing to other coarse-grained levels. ICT proposes new quantitative definitions of both conscious content and conscious level. With the parsimonious definitions and a hypothesise, ICT provides explanations and predictions of various phenomena associated with consciousness. The implications of ICT naturally reconciles issues in many existing theories of consciousness and provides explanations for many of our intuitions about consciousness. Most importantly, ICT demonstrates that information can be the common language between consciousness and physical reality.

q-bio.NC↗

arXiv2018

A unified strategy for implementing curiosity and empowerment driven reinforcement learning

Although there are many approaches to implement intrinsically motivated artificial agents, the combined usage of multiple intrinsic drives remains still a relatively unexplored research area. Specifically, we hypothesize that a mechanism capable of quantifying and controlling the evolution of the information flow between the agent and the environment could be the fundamental component for implementing a higher degree of autonomy into artificial intelligent agents. This paper propose a unified strategy for implementing two semantically orthogonal intrinsic motivations: curiosity and empowerment. Curiosity reward informs the agent about the relevance of a recent agent action, whereas empowerment is implemented as the opposite information flow from the agent to the environment that quantifies the agent's potential of controlling its own future. We show that an additional homeostatic drive is derived from the curiosity reward, which generalizes and enhances the information gain of a classical curious/heterostatic reinforcement learning agent. We show how a shared internal model by curiosity and empowerment facilitates a more efficient training of the empowerment function. Finally, we discuss future directions for further leveraging the interplay between these two intrinsic rewards.

cs.AI↗

arXiv2018

Boredom-driven curious learning by Homeo-Heterostatic Value Gradients

This paper presents the Homeo-Heterostatic Value Gradients (HHVG) algorithm as a formal account on the constructive interplay between boredom and curiosity which gives rise to effective exploration and superior forward model learning. We envisaged actions as instrumental in agent's own epistemic disclosure. This motivated two central algorithmic ingredients: devaluation and devaluation progress, both underpin agent's cognition concerning intrinsically generated rewards. The two serve as an instantiation of homeostatic and heterostatic intrinsic motivation. A key insight from our algorithm is that the two seemingly opposite motivations can be reconciled---without which exploration and information-gathering cannot be effectively carried out. We supported this claim with empirical evidence, showing that boredom-enabled agents consistently outperformed other curious or explorative agent variants in model building benchmarks based on self-assisted experience accumulation.

cs.AI↗

arXiv2018

Being curious about the answers to questions: novelty search with learned attention

We investigate the use of attentional neural network layers in order to learn a `behavior characterization' which can be used to drive novelty search and curiosity-based policies. The space is structured towards answering a particular distribution of questions, which are used in a supervised way to train the attentional neural network. We find that in a 2d exploration task, the structure of the space successfully encodes local sensory-motor contingencies such that even a greedy local `do the most novel action' policy with no reinforcement learning or evolution can explore the space quickly. We also apply this to a high/low number guessing game task, and find that guessing according to the learned attention profile performs active inference and can discover the correct number more quickly than an exact but passive approach.

cs.AI↗

arXiv2018

Learning to generate classifiers

We train a network to generate mappings between training sets and classification policies (a 'classifier generator') by conditioning on the entire training set via an attentional mechanism. The network is directly optimized for test set performance on an training set of related tasks, which is then transferred to unseen 'test' tasks. We use this to optimize for performance in the low-data and unsupervised learning regimes, and obtain significantly better performance in the 10-50 datapoint regime than support vector classifiers, random forests, XGBoost, and k-nearest neighbors on a range of small datasets.

cs.LG↗

arXiv2018

Efficient Algorithms for Searching the Minimum Information Partition in Integrated Information Theory

The ability to integrate information in the brain is considered to be an essential property for cognition and consciousness. Integrated Information Theory (IIT) hypothesizes that the amount of integrated information ($Φ$) in the brain is related to the level of consciousness. IIT proposes that to quantify information integration in a system as a whole, integrated information should be measured across the partition of the system at which information loss caused by partitioning is minimized, called the Minimum Information Partition (MIP). The computational cost for exhaustively searching for the MIP grows exponentially with system size, making it difficult to apply IIT to real neural data. It has been previously shown that if a measure of $Φ$ satisfies a mathematical property, submodularity, the MIP can be found in a polynomial order by an optimization algorithm. However, although the first version of $Φ$ is submodular, the later versions are not. In this study, we empirically explore to what extent the algorithm can be applied to the non-submodular measures of $Φ$ by evaluating the accuracy of the algorithm in simulated data and real neural data. We find that the algorithm identifies the MIP in a nearly perfect manner even for the non-submodular measures. Our results show that the algorithm allows us to measure $Φ$ in large systems within a practical amount of time.

q-bio.NC↗

arXiv2018

Curiosity-driven reinforcement learning with homeostatic regulation

We propose a curiosity reward based on information theory principles and consistent with the animal instinct to maintain certain critical parameters within a bounded range. Our experimental validation shows the added value of the additional homeostatic drive to enhance the overall information gain of a reinforcement learning agent interacting with a complex environment using continuous actions. Our method builds upon two ideas: i) To take advantage of a new Bellman-like equation of information gain and ii) to simplify the computation of the local rewards by avoiding the approximation of complex distributions over continuous states and actions.

cs.AI↗

arXiv2017

Learning body-affordances to simplify action spaces

Controlling embodied agents with many actuated degrees of freedom is a challenging task. We propose a method that can discover and interpolate between context dependent high-level actions or body-affordances. These provide an abstract, low-dimensional interface indexing high-dimensional and time- extended action policies. Our method is related to recent ap- proaches in the machine learning literature but is conceptually simpler and easier to implement. More specifically our method requires the choice of a n-dimensional target sensor space that is endowed with a distance metric. The method then learns an also n-dimensional embedding of possibly reactive body-affordances that spread as far as possible throughout the target sensor space.

cs.AI↗

arXiv2017

Counterfactual Control for Free from Generative Models

We introduce a method by which a generative model learning the joint distribution between actions and future states can be used to automatically infer a control scheme for any desired reward function, which may be altered on the fly without retraining the model. In this method, the problem of action selection is reduced to one of gradient descent on the latent space of the generative model, with the model itself providing the means of evaluating outcomes and finding the gradient, much like how the reward network in Deep Q-Networks (DQN) provides gradient information for the action generator. Unlike DQN or Actor-Critic, which are conditional models for a specific reward, using a generative model of the full joint distribution permits the reward to be changed on the fly. In addition, the generated futures can be inspected to gain insight in to what the network 'thinks' will happen, and to what went wrong when the outcomes deviate from prediction.

cs.LG↗

arXiv2017

A description length approach to determining the number of k-means clusters

We present an asymptotic criterion to determine the optimal number of clusters in k-means. We consider k-means as data compression, and propose to adopt the number of clusters that minimizes the estimated description length after compression. Here we report two types of compression ratio based on two ways to quantify the description length of data after compression. This approach further offers a way to evaluate whether clusters obtained with k-means have a hierarchical structure by examining whether multi-stage compression can further reduce the description length. We applied our criteria to determine the number of clusters to synthetic data and empirical neuroimaging data to observe the behavior of the criteria across different types of data set and suitability of the two types of criteria for different datasets. We found that our method can offer reasonable clustering results that are useful for dimension reduction. While our numerical results revealed dependency of our criteria on the various aspects of dataset such as the dimensionality, the description length approach proposed here provides a useful guidance to determine the number of clusters in a principled manner when underlying properties of the data are unknown and only inferred from observation of data.

stat.ML↗

arXiv2017

Integrated information and dimensionality in continuous attractor dynamics

There has been increasing interest in the integrated information theory (IIT) ofconsciousness, which hypothesizes that consciousness is integrated information withinneuronal dynamics. However, the current formulation of IIT poses both practical andtheoretical problems when we aim to empirically test the theory by computingintegrated information from neuronal signals. For example, measuring integratedinformation requires observing all the elements in the considered system at the sametime, but this is practically rather difficult. In addition, the interpretation of the spatialpartition needed to compute integrated information becomes vague in continuous time-series variables due to a general property of nonlinear dynamical systems known as"embedding." Here, we propose that some aspects of such problems are resolved byconsidering the topological dimensionality of shared attractor dynamics as an indicatorof integrated information in continuous attractor dynamics. In this formulation, theeffects of unobserved nodes on the attractor dynamics can be reconstructed using atechnique called delay embedding, which allows us to identify the dimensionality of anembedded attractor from partial observations. We propose that the topologicaldimensionality represents a critical property of integrated information, as it is invariantto general coordinate transformations. We illustrate this new framework with simpleexamples and discuss how it fits together with recent findings based on neuralrecordings from awake and anesthetized animals. This topological approach extendsthe existing notions of IIT to continuous dynamical systems and offers a much-neededframework for testing the theory with experimental data by substantially relaxing theconditions required for evaluating integrated information in real neural systems.

q-bio.NC↗

arXiv2016

Permutation-equivariant neural networks applied to dynamics prediction

The introduction of convolutional layers greatly advanced the performance of neural networks on image tasks due to innately capturing a way of encoding and learning translation-invariant operations, matching one of the underlying symmetries of the image domain. In comparison, there are a number of problems in which there are a number of different inputs which are all 'of the same type' --- multiple particles, multiple agents, multiple stock prices, etc. The corresponding symmetry to this is permutation symmetry, in that the algorithm should not depend on the specific ordering of the input data. We discuss a permutation-invariant neural network layer in analogy to convolutional layers, and show the ability of this architecture to learn to predict the motion of a variable number of interacting hard discs in 2D. In the same way that convolutional layers can generalize to different image sizes, the permutation layer we describe generalizes to different numbers of objects.

cs.CV↗

arXiv2016

Neural Coarse-Graining: Extracting slowly-varying latent degrees of freedom with neural networks

We present a loss function for neural networks that encompasses an idea of trivial versus non-trivial predictions, such that the network jointly determines its own prediction goals and learns to satisfy them. This permits the network to choose sub-sets of a problem which are most amenable to its abilities to focus on solving, while discarding 'distracting' elements that interfere with its learning. To do this, the network first transforms the raw data into a higher-level categorical representation, and then trains a predictor from that new time series to its future. To prevent a trivial solution of mapping the signal to zero, we introduce a measure of non-triviality via a contrast between the prediction error of the learned model with a naive model of the overall signal statistics. The transform can learn to discard uninformative and unpredictable components of the signal in favor of the features which are both highly predictive and highly predictable. This creates a coarse-grained model of the time-series dynamics, focusing on predicting the slowly varying latent parameters which control the statistics of the time-series, rather than predicting the fast details directly. The result is a semi-supervised algorithm which is capable of extracting latent parameters, segmenting sections of time-series with differing statistics, and building a higher-level representation of the underlying dynamics from unlabeled data.

cs.AI↗