Mimicking the brain: Deep learning meets vector-symbolic AI

symbolic machine learning

They involve every individual memory entry instead of a single discrete entry. McCarthy’s approach to fix the frame problem was circumscription, a kind of non-monotonic logic where deductions could be made from actions that need only specify what would change while not having to explicitly specify everything that would not change. Other non-monotonic logics provided truth maintenance systems that revised beliefs leading to contradictions.

symbolic machine learning

Second, the participants responded to the query instructions all at once, on a single web page, allowing the participants to edit, go back and forth, and responses. By contrast, the previous experiment collected the query responses one by one and had a curriculum of multiple distinct stages of learning. By providing a table of this data to a symbolic regression engine, it will start randomly trying different combinations of the input features and the base functions to try to predict the target variable while keeping track of the best formulas found so far. The CGBE strategies described in “Strategies for Learning Tree-to-Tree Mappings” are oriented towards the situation of refinement of a source language to a target language, such as the code generation case. In this situation, the source syntax trees are typically less elaborately structured than the target trees.


The advantage of neural networks is that they can deal with messy and unstructured data. Instead of manually laboring through the rules of detecting cat pixels, you can train a deep learning algorithm on many pictures of cats. When you provide it with a new image, it will return the probability that it contains a cat. On the other side of the spectrum of model complexity are black-box regressors like neural networks, which transform the input data through a series of implicit calculations before giving a result. Those models are very popular nowadays due to the promise that they will one day result in a general “artificial intelligence”, and due to their striking success in difficult problems like computer vision.

  • This approach may not be applicable in cases where the source and target languages have a large syntactic distance, such as a 3GL and assembly language.
  • Finally, their operation is largely opaque to humans, rendering them unsuitable for domains in which verifiability is important.
  • The store could act as a knowledge base and the clauses could act as rules or a restricted form of logic.
  • Cognitive architectures such as ACT-R may have additional capabilities, such as the ability to compile frequently used knowledge into higher-level chunks.

We would like to emphasize this since so far as machine learning has only been applied with objective attributes gained from image analysis, which rather correspond to formal-perceptive attributes18,19,20,21. According to our results, however, humans appear to judge artwork’s creativity according to the level of content and aspects of meaning. We predicted creativity judgement ratings of artworks based on ratings of 17 visual art-attributes using machine learning. We computed multivariate Random Forest (RF) regressor models and determined their prediction performances. On average the RF model’s predicting creativity judgements were off by 17.5 points from the actual judgements, within a creativity judgement range of 1 to 101 (see Table 2). The coefficient of determination (R2) between the predicted creativity judgements and the actual creativity judgements was on average 0.30, which indicates that the RF models explain on average 30% of the total variance in creativity judgement ratings.

Symbolic machine learning improved MCFT model for punching shear resistance of FRP-reinforced concrete slabs

This leaves open questions and implications for future research seeking to deepening the understanding of the concept of emotionality as creative expression in art itself. The analysis of our data uncovered an interplay among the relevant predictors, i.e., the attributes, and the target variable, creativity. Further analysis revealed, that while no significant correlation was found between the predictors and creativity, there were some observable correlations among the predictors themselves. This pattern is depicted in a correlation heatmap (see Fig. S3 Supplementary Information). Specifically, the correlation coefficients between symbolism and emotionality, and between imaginativeness and emotionality were low, with values of 0.22 and 0.16 respectively.


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