Human-like systematic generalization through a meta-learning neural network

Last, function 3 (‘kiki’) takes both the preceding and following strings as input, processes them and concatenates their outputs in reverse order (‘dax kiki lug’ is BLUE RED). We also tested function 3 in cases in which its arguments were generated by the other functions, exploring function composition (‘wif blicket dax kiki lug’ is BLUE GREEN RED GREEN). During the study phase (see description below), participants saw examples that disambiguated the order of function application for the tested compositions (function 3 takes scope over the other functions). Over 35 years ago, when Fodor and Pylyshyn raised the issue of systematicity in neural networks1, today’s models19 and their language skills were probably unimaginable. As a credit to Fodor and Pylyshyn’s prescience, the systematicity debate has endured.

For example, once a child learns how to ‘skip’, they can understand how to ‘skip backwards’ or ‘skip around a cone twice’ due to their compositional skills. Fodor and Pylyshyn1 argued that neural networks lack this type of systematicity and are therefore not plausible cognitive models, leading to a vigorous debate that spans 35 years2,3,4,5. The first is that human compositional skills, although important, may not be as systematic and rule-like as Fodor and Pylyshyn indicated3,6,7. The second is that neural networks, although limited in their most basic forms, can be more systematic when using sophisticated architectures8,9,10. In recent years, neural networks have advanced considerably and led to a number of breakthroughs, including in natural language processing.

The current state of symbolic AI

In contrast, because the rules we learn are symbolic, they can be independent of the specific input data (such as variable names) and apply regardless of the size of these data. To address this gap, we conducted a study using 17 visual art-attributes assessed through semantic differential scales45. These attributes, such as warm/cold, simple/complex, emotionless/emotionally loaded46, have been identified in art research as influential factors in rating artworks14,28,47,48,49,50. We employed Random Forest machine learning regression models, which can learn non-linear association patterns and interactions from data39, to predict creativity judgments based on the aforementioned attributes. To analyze the importance of each individual art-attribute in predicting creativity judgments, we utilized permutation importance, a method from the interpretable machine learning field6.

Machine learning (ML) can be defined as a technology which aims to automate the learning of knowledge from the instances of a training set, such that the learned knowledge can then be applied to derive information about other instances. A wide range of techniques are encompassed by this definition, including statistical approaches, traditional and recurrent neural networks, and symbolic learning approaches such as decision trees [29] or inductive logic programming (ILP) [26]. Machine learning typically uses a training phase, during which knowledge is induced from the training set, and a validation phase, where the accuracy of the learned knowledge is tested against a new dataset, for which the expected results are known. Admittedly, there are empirical models used for evaluating the punching shear resistance of FRP-reinforced concrete slabs, an inherent problem cannot be avoided that the oversimplifications may occur in the derivation of theoretical models [18].

Code, Data and Media Associated with this Article

Each method executes a series of rule-based instructions that might read and change the properties of the current and other objects. Despite the efforts to promote symbolic regression over the years, the truth is that this method has never gained mainstream popularity. In an academic context, research on hot topics like neural networks is much more tractable, given that optimal algorithms are known for training the model.

Read more about https://www.metadialog.com/ here.