scholarly journals Bayesian Models of Conceptual Development: Learning as Building Models of the World

2020 ◽  
Author(s):  
Tomer David Ullman ◽  
Joshua Tenenbaum

A Bayesian framework helps to address, in computational terms, what knowledgechildren start with and how they construct and adapt models of the worldduring childhood. Within this framework, inference over hierarchies of probabilisticgenerative programs in particular offers a normative and descriptiveaccount of children's model-building. We consider two classic settings in whichcognitive development has been framed as model-building: (i) Core knowledgein infancy, and (ii) The child as scientist. We interpret learning in both of thesesettings as resource-constrained, hierarchical Bayesian program induction withdifferent primitives and constraints. We examine what mechanisms childrencould use to meet the algorithmic challenges of navigating large spaces of potentialmodels, in particular the proposal of \the child as hacker" and how itmight be realized drawing on recent computational advances. We also discussprospects for a unifying account of model building across scientific theories andintuitive theories, and in biological and cultural evolution more generally.

2020 ◽  
Vol 2 (1) ◽  
pp. 533-558
Author(s):  
Tomer D. Ullman ◽  
Joshua B. Tenenbaum

A Bayesian framework helps address, in computational terms, what knowledge children start with and how they construct and adapt models of the world during childhood. Within this framework, inference over hierarchies of probabilistic generative programs in particular offers a normative and descriptive account of children's model building. We consider two classic settings in which cognitive development has been framed as model building: ( a) core knowledge in infancy and ( b) the child as scientist. We interpret learning in both of these settings as resource-constrained, hierarchical Bayesian program induction with different primitives and constraints. We examine what mechanisms children could use to meet the algorithmic challenges of navigating large spaces of potential models, in particular the proposal of the child as hacker and how it might be realized by drawing on recent computational advances. We also discuss prospects for a unifying account of model building across scientific theories and intuitive theories, and in biological and cultural evolution more generally.


Author(s):  
N. Thompson Hobbs ◽  
Mevin B. Hooten

This chapter lays out the basic principles of Bayesian inference, building on the concepts of probability developed in Chapter 3. It seeks to use the rules of probability to show how Bayes' theorem works, by making use of the conditional rule of probability and the law of total probability. The chapter begins with the central, underpinning tenet of the Bayesian view: the world can be divided into quantities that are observed and quantities that are unobserved. Unobserved quantities include parameters in models, latent states predicted by models, missing data, effect sizes, future states, and data before they are observed. We wish to learn about these quantities using observations. The Bayesian framework for achieving that understanding is applied in exactly the same way regardless of the specifics of the research problem at hand or the nature of the unobserved quantities.


Author(s):  
Ruth Garrett Millikan

This book weaves together themes from natural ontology, philosophy of mind, philosophy of language and information, areas of inquiry that have not recently been treated together. The sprawling topic is Kant’s how is knowledge possible? but viewed from a contemporary naturalist standpoint. The assumption is that we are evolved creatures that use cognition as a guide in dealing with the natural world, and that the natural world is roughly as natural science has tried to describe it. Very unlike Kant, then, we must begin with ontology, with a rough understanding of what the world is like prior to cognition, only later developing theories about the nature of cognition within that world and how it manages to reflect the rest of nature. And in trying to get from ontology to cognition we must traverse another non-Kantian domain: questions about the transmission of information both through natural signs and through purposeful signs including, especially, language. Novelties are the introduction of unitrackers and unicepts whose job is to recognize the same again as manifested through the jargon of experience, a direct reference theory for common nouns and other extensional terms, a naturalist sketch of uniceptual—roughly conceptual— development, a theory of natural information and of language function that shows how properly functioning language carries natural information, a novel description of the semantics/pragmatics distinction, a discussion of perception as translation from natural informational signs, new descriptions of indexicals and demonstratives and of intensional contexts and a new analysis of the reference of incomplete descriptions.


2010 ◽  
Vol 18 (3) ◽  
pp. 329-345
Author(s):  
Hubert Markl

The reason why I wavered a bit with this topic is that, after all, it has to do with Darwin, after a great Darwin year, as seen by a German scientist. Not that Darwin was very adept in German: Gregor Mendel’s ‘Versuche über Pflanzenhybriden’ (Experiments on Plant Hybrids) was said to have stayed uncut and probably unread on his shelf, which is why he never got it right with heredity in his life – only Gregory Bateson, Ronald A. Fisher, and JBS Haldane, together with Sewall Wright merged evolution with genetics. But Darwin taught us, nevertheless, in essence why the single human species shows such tremendous ethnic diversity, which impresses us above all through a diversity of languages – up to 7000 altogether – and among them, as a consequence, also German, my mother tongue, and English. It would thus have been a truly Darwinian message, if I had written this article in German. I would have called that the discommunication function of the many different languages in humans, which would have been a most significant message of cultural evolution, indeed. I finally decided to overcome the desire to demonstrate so bluntly what cultural evolution is all about, or rather to show that nowadays, with global cultural progress, ‘the world is flat’ indeed – even linguistically. The real sign of its ‘flatness’ is that English is used everywhere, even if Thomas L. Friedman may not have noticed this sign. But I will also come back to that later, when I hope to show how Darwinian principles connect both natural and cultural evolution, and how they first have been widely misunderstood as to their true meaning, and then have been terribly misused – although more so by culturalists, or some self-proclaimed ‘humanists’, rather than by biologists – or at least most of them. Let me, however, quickly add a remark on human languages. That languages even influence our brains and our thinking, that is: how we see the world, has first been remarked upon by Wilhelm von Humboldt and later, more extensively so, by Benjamin Whorf. It has recently been shown by neural imaging – for instance by Angela Friederici – that one’s native language, first as learned from one’s mother and from those around us when we are babies, later from one’s community of speakers, can deeply impinge on a baby’s brain development and stay imprinted in it throughout life, even if language is, of course, learned and not fully genetically preformed. This shows once more how deep the biological roots are that ground our cultures, according to truly Darwinian principles, even if these cultures are completely learned.


2014 ◽  
Vol 61 (1) ◽  
pp. 116-132 ◽  
Author(s):  
Xi Li ◽  
Kiwamu Ishikura ◽  
Chunying Wang ◽  
Jagadeesh Yeluripati ◽  
Ryusuke Hatano

2021 ◽  
Vol 5 (4) ◽  
pp. 1-28
Author(s):  
Chia-Heng Tu ◽  
Qihui Sun ◽  
Hsiao-Hsuan Chang

Monitoring environmental conditions is an important application of cyber-physical systems. Typically, the monitoring is to perceive surrounding environments with battery-powered, tiny devices deployed in the field. While deep learning-based methods, especially the convolutional neural networks (CNNs), are promising approaches to enriching the functionalities offered by the tiny devices, they demand more computation and memory resources, which makes these methods difficult to be adopted on such devices. In this article, we develop a software framework, RAP , that permits the construction of the CNN designs by aggregating the existing, lightweight CNN layers, which are able to fit in the limited memory (e.g., several KBs of SRAM) on the resource-constrained devices satisfying application-specific timing constrains. RAP leverages the Python-based neural network framework Chainer to build the CNNs by mounting the C/C++ implementations of the lightweight layers, trains the built CNN models as the ordinary model-training procedure in Chainer, and generates the C version codes of the trained models. The generated programs are compiled into target machine executables for the on-device inferences. With the vigorous development of lightweight CNNs, such as binarized neural networks with binary weights and activations, RAP facilitates the model building process for the resource-constrained devices by allowing them to alter, debug, and evaluate the CNN designs over the C/C++ implementation of the lightweight CNN layers. We have prototyped the RAP framework and built two environmental monitoring applications for protecting endangered species using image- and acoustic-based monitoring methods. Our results show that the built model consumes less than 0.5 KB of SRAM for buffering the runtime data required by the model inference while achieving up to 93% of accuracy for the acoustic monitoring with less than one second of inference time on the TI 16-bit microcontroller platform.


Author(s):  
P. Kyle Stanford

This chapter seeks to explore and develop the proposal that even our best scientific theories are not (as the scientific realist would have it) accurate descriptions of how things stand in otherwise inaccessible domains of nature but are instead simply powerful conceptual tools or instruments for engaging practically with the world around us. It describes a number of persistent challenges facing any attempt to apply the American Pragmatists’ global conception of all ideas, beliefs, theories, and cognitions quite generally as such tools or instruments to only a restricted class or category of such entities (such as our best scientific theories) instead. It then seeks to overcome these challenges by regarding scientific instrumentalism as simply applying the scientific realist’s own attitude toward a theory like Newtonian mechanics to even the most empirically successful and instrumentally powerful theory we have in any given scientific domain.


Sign in / Sign up

Export Citation Format

Share Document