Stochastic parallel extreme artificial hydrocarbon networks: An implementation for fast and robust supervised machine learning in high-dimensional data

Estimation of Low Nutrients in Tomato Crops Through the Analysis of Leaf Images Using Machine Learning

“…Its key feature can be described as the ability to package data in units of so-called molecules. Then, packages are organized and optimized through heuristic mechanisms based on chemical assumptions that are encoded in the training algorithm [22].…”

“…Those are structures that represent nonlinearities among molecules. They are associated with a functional behavior as in (2), where m is the number of molecules in the compound and Σ j is a partition of the input x such that Σ j = {x| arg min j (x − µ j ) = j}, and µ j ∈ R n is the center of the jth molecule [22]. In fact, Σ j1 ∩ Σ j2 = ∅ if j 1 = j 2 .…”

“…They differ in terms of how to approach the learning process of the molecular parameters and the centers of molecules. In this work, we adopted the stochastic parallel extreme (SPE-AHN) training algorithm, and further details can be consulted in [22].…”

“…Literature reports that large number of features in data might reduce the predictability power of the AHN [22]. To minimize the impact of large number of features from the last convolutional layer, we propose to implement a feature reduction layer in the CNN+AHN before the AHN.…”

“…The architecture of this CNN+AHN model comprises a set of convolutional layers, as feature extraction, and an artificial hydrocarbon network (AHN) model as the dense layer. Roughly speaking, AHN is a supervised learning method [21] that models data using carbon networks as inspiration that promotes modular organization of data, structural stability of data-packages, and inheritance of packaging information [22].…”