Systems and methods for generating hash trees and using neural networks to process the same
View Patent ↗The present disclosure relates to systems and methods for generating a persona and detecting anomalies using a hash tree. In one implementation, the system may include one or more memories storing instructions and one or more processors configured to execute the instructions. The instructions may include instructions to receive a plurality of data structures related to an individual, convert the data structures into a plurality of Bayesian wavelets, group the Bayesian wavelets into a tree structure, using transitions between the Bayesian wavelets within the tree structure, generate a plurality of Markovian wavelets representing the transitions, replace one or more of the Bayesian wavelets with hashes, and output the tree structure as a persona representing the individual. The instructions may also include training a neural network.
1. A system for training a deep field network to detect anomalies within a hash tree, the system comprising:
one or more memories storing instructions; and
one or more processors configured to execute the instructions to:
receive a plurality of tree structures representing individuals, each tree structure including Bayesian wavelets and Markovian wavelets governing transitions between the Bayesian wavelets and being configured as a persona representing a corresponding individual,
replace one or more of the Bayesian wavelets with hashes of the one or more Bayesian wavelets;
group the Bayesian wavelets in the tree structures by coefficients,
convert the Bayesian wavelets to tensors;
train, using the tensors converted from the Bayesian wavelets, a neural network for each group independently of other groups while increasing a likelihood of a corresponding wavelet being anomalistic within each group,
integrate the neural networks into a deep field network, wherein combining a new wavelet with the neural networks amplifies a first neural network corresponding to a first group having a first set of coefficients and minimizes a second neural network corresponding to a second group having a second set of coefficients,
calculate a difference field between a corresponding tensor representing the new wavelet and a corresponding field representing the tensors from the tree structure to form a difference vector, and
detect one or more anomalies within the tree structures using the integrated neural networks by determining whether a corresponding tensor is anomalous based on one or more thresholds applied to the difference vector, wherein the one or more anomalies indicate an alteration in harmonics of the tree structures.
2. The system of claim 1 , wherein each neural network is further trained specific to at least one of a particular tree structure, a particular location, or a particular time of day.
3. The system of claim 1 , wherein converting the wavelets to tensors includes applying a discrete wavelet transform.
4. The system of claim 3 , wherein the discrete wavelet transform is performed using a filter bank having a plurality of convolutional-accumulators.
5. The system of claim 4 , wherein the convolutional accumulators are configured to accumulate using base e.
6. The system of claim 4 , wherein the discrete wavelet transform includes an exponential smoothing average in the filter bank.
7. The system of claim 4 , wherein the one or more processors further execute the instructions to convert the Bayesian wavelets to tensors that are output with the tree structures.