Anomaly analysis using a blockchain, and applications thereof

Abstract

Disclosed herein are system, method, and computer program product embodiments for scrubbing anomalies from an expanding dataset. In an embodiment, a data sanitization system may determine whether data is anomalous to a set of data stored on a first blockchain. The data sanitization system may perform this determination using a first machine learning algorithm trained using the set of data. Upon determining that data is anomalous, the data sanitization system may publish the data in a second blockchain different from the first blockchain. The data sanitization system may monitor data of the second blockchain and apply a second machine learning algorithm to this data to identify a pattern of anomalous data. In response to identifying the pattern, the data sanitization system may publish the anomalous data of the second blockchain to the first blockchain.

What is claimed is: 
     
         1 . A computer-implemented method for scrubbing anomalies from an expanding dataset, comprising:
 training a first machine learning algorithm using a set of data published on a sanitized blockchain;   applying the first machine learning algorithm to a first data to designate the first data as outlier data relative to the set of data in the sanitized blockchain;   publishing the first data to an anomaly blockchain;   applying a second machine learning algorithm to data stored in the anomaly blockchain to identify a pattern of anomalous data; and   publishing the anomalous data from the anomaly blockchain to the sanitized blockchain.   
     
     
         2 . The computer-implemented method of  claim 1 , further comprising:
 receiving a second data;   applying the first machine learning algorithm to the second data to determine whether the second data conforms to the set of data in the sanitized blockchain;   publishing the second data to the sanitized blockchain to generate an updated set of data; and   retraining the first machine learning algorithm using the updated set of data.   
     
     
         3 . The computer-implemented method of  claim 2 , wherein the retraining occurs using an unsupervised learning technique. 
     
     
         4 . The computer-implemented method of  claim 1 , wherein applying the first machine learning algorithm further comprises:
 applying a clustering technique to the first data and the set of data in the sanitized blockchain; and   designating the first data as outlier data by identifying the first data as being outside of a cluster determined from the set of data in the sanitized blockchain.   
     
     
         5 . The computer-implemented method of  claim 1 , wherein applying the second machine learning algorithm further comprises:
 determining the data stored in the anomaly blockchain exceeds a threshold amount; and   when the data stored in the anomaly blockchain exceeds the threshold amount, applying a clustering technique to the data stored in the anomaly blockchain; and   determining the data stored in the anomaly blockchain forms a grouped pattern.   
     
     
         6 . The computer-implemented method of  claim 1 , further comprising:
 training a third machine learning algorithm using the set of data in the sanitized blockchain to provide a predictive analysis in response to user queries.   receiving, from a user device, a query including a parameter value corresponding to a field in the set of data published on the sanitized blockchain; and   returning, to the user device, a prediction generated by the third machine learning algorithm based on the parameter value and the set of data on the sanitized blockchain.   
     
     
         7 . The computer-implemented method of  claim 1 , wherein the set of data published on the sanitized blockchain includes transaction data related to vehicle transactions. 
     
     
         8 . The computer-implemented method of  claim 6 , wherein the transaction data related to vehicle transactions further comprises at least one of the following: make and model of a vehicle, year of the vehicle, mileage of the vehicle, a geographical location corresponding to a sale, and price of the sale. 
     
     
         9 . A system for scrubbing anomalies from an expanding dataset, comprising:
 a memory; and   at least one processor coupled to the memory and configured to:
 train a first machine learning algorithm using a set of data published on a sanitized blockchain; 
 apply the first machine learning algorithm to a first transaction data to designate the first transaction data as outlier data relative to the set of data in the sanitized blockchain; 
 publish the first transaction data to an anomaly blockchain; 
 apply a second machine learning algorithm to data stored in the anomaly blockchain to identify a pattern of anomalous data; and 
 publish the anomalous data from the anomaly blockchain to the sanitized blockchain. 
   
     
     
         10 . The system of  claim 9 , wherein the at least one processor is further configured to:
 receive a second transaction data;   apply the first machine learning algorithm to the second transaction data to determine whether the second transaction data conforms to the set of data in the sanitized blockchain;   publish the second transaction data to the sanitized blockchain to generate an updated set of data; and   retrain the first machine learning algorithm using the updated set of data.   
     
     
         11 . The system of  claim 10 , wherein the retraining occurs using an unsupervised learning technique. 
     
     
         12 . The system of  claim 9 , wherein to apply the first machine learning algorithm, the at least one processor is further configured to:
 apply a clustering technique to the first transaction data and the set of data in the sanitized blockchain; and   designate the first transaction data as outlier data by identifying the first transaction data as being outside of a cluster determined from the set of data published in the sanitized blockchain.   
     
     
         13 . The system of  claim 9 , wherein to apply the second machine learning algorithm, the at least one processor is further configured to:
 determine the data stored in the anomaly blockchain exceeds a threshold amount;   when the data stored in the anomaly blockchain exceeds the threshold amount, apply a clustering technique to the data stored in the anomaly blockchain; and   determine the data stored in the anomaly blockchain forms a grouped pattern.   
     
     
         14 . The system of  claim 9 , wherein the at least one processor is further configured to:
 train a third machine learning algorithm using the set of data in the sanitized blockchain to provide a predictive analysis in response to user queries.   receive, from a user device, a query including a parameter value corresponding to a field in the set of data published on the sanitized blockchain; and   return, to the user device, a prediction generated by the third machine learning algorithm based on the parameter value and the set of data on the sanitized blockchain.   
     
     
         15 . A non-transitory computer-readable device having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:
 applying a first machine learning algorithm to a first data to determine whether the first data should be designated as outlier data relative to a set of data used to train the first machine learning algorithm, wherein the set of data is published on a sanitized blockchain;   publishing the first data to an anomaly blockchain;   applying a second machine learning algorithm to data stored in the anomaly blockchain to identify a pattern of anomalous data;   publishing the anomalous data from the anomaly blockchain to the sanitized blockchain;   training a third machine learning algorithm using the set of data on the sanitized blockchain to provide a predictive analysis in response to user queries;   receiving, from a user device, a query including a parameter value corresponding to a field in the set of data on the sanitized blockchain; and   returning, to the user device, a prediction generated by the third machine learning algorithm based on the parameter value and the set of data on the sanitized blockchain.   
     
     
         16 . The non-transitory computer-readable device of  claim 15 , the operations further comprising:
 receiving a second data;   applying the first machine learning algorithm to the second data to determine whether the second data conforms to the set of data published on the sanitized blockchain;   publishing the second data to the sanitized blockchain to generate an updated set of data; and   retraining the first machine learning algorithm and the third machine learning algorithm using the updated set of data.   
     
     
         17 . The non-transitory computer-readable device of  claim 16 , wherein the retraining occurs using an unsupervised learning technique. 
     
     
         18 . The non-transitory computer-readable device of  claim 15 , wherein applying the first machine learning algorithm further comprises:
 applying a clustering technique to the first data and the set of data published on the sanitized blockchain; and   determining the first data to be anomalous by identifying the first data as being outside of a cluster determined from the set of data published on the sanitized blockchain.   
     
     
         19 . The non-transitory computer-readable device of  claim 15 , wherein applying the second machine learning algorithm further comprises:
 determining the data stored in the anomaly blockchain exceeds a threshold amount; and   when the data stored in the anomaly blockchain exceeds the threshold amount, applying a clustering technique to the data stored in the anomaly blockchain; and   determining the data stored in the anomaly blockchain forms a grouped pattern.   
     
     
         20 . The non-transitory computer-readable device of  claim 15 , wherein the set of data includes data related to vehicle transactions.