Georgia Tech Research Corporation

Symantec Corporation

University of Pennsylvania

Ca, Inc.

Nortonlifelock Inc.

Javier Echauz

Brian Litt

Andrew B Gardner

Rosana Esteller

Saurabh Shintre

George J Vachtsevanos

Gregory A Worrell

Reuben Feinman

Jasjeet Dhaliwal

Sarfaraz Hussein

John Keith Kenemer

George J Vachtesvanos

Keith Kenemer

Peng Wang

Muid Mufti

Lewis J Dorrity

Machine learning adversarial campaign mitigation on a computing device. The method may include deploying an original machine learning model in a model environment associated with a client device; deploying a classification monitor in the model environment to monitor classification decision outputs in the machine learning model; detecting, by the classification monitor, a campaign of adversarial classification decision outputs in the machine learning model; applying a transformation function to the machine learning model in the model environment to transform the adversarial classification decision outputs to thwart the campaign of adversarial classification decision outputs; determining a malicious attack on the client device based in part on detecting the campaign of adversarial classification decision outputs; and implementing a security action to protect the computing device against the malicious attack.

Machine learning adversarial campaign mitigation on a computing device

The disclosed computer-implemented method for adaptively managing data drift in a classifier may include (i) receiving, at a computing device, an input sample of digital information having an unknown reputation and (ii) performing a security action that may include (A) identifying the input sample as benign or malicious based on a result obtained by classifying the input sample using a machine learning model trained using activity regularization, (B) calculating an internal activity of the machine learning model occurring during the classifying, (C) calculating an activation entropy of the machine learning model occurring during the classifying, (D) comparing a combination of the internal activity and the activation entropy to a threshold, and (E) when the combination of the internal activity and the activation entropy meets or exceeds the threshold, identifying the result as a low-confidence result. Various other methods, systems, and computer-readable media are also disclosed.

Systems and methods for adaptively managing data drift in a classifier

The disclosed computer-implemented method for trichotomous malware classification may include (1) identifying a sample potentially representing malware, (2) selecting a machine learning model trained on a set of samples to distinguish between malware samples and benign samples, (3) analyzing the sample using a plurality of stochastically altered versions of the machine learning model to produce a plurality of classification results, (4) calculating a variance of the plurality of classification results, and (5) classifying the sample based at least in part on the variance of the plurality of classification results. Various other methods, systems, and computer-readable media are also disclosed.

Systems and methods for trichotomous malware classification

Developing a measure of critical systems-like behavior in an epilepsy patient in order to map epileptic networks, either passively or evoking responses through subthreshold stimulation, and to apply 'therapeutic' stimulations to the patient that cause smaller, but more frequent dissipations of 'energy,' a transcription product, subclinical electrophysiological activity or seizures in order to raise the clinical seizure initiation threshold, through releasing accumulated interictal energy in a seizure onset zone or elsewhere in the epileptic network, thereby preventing occurrence of larger more debilitating seizures.

Active control of epileptic seizures and diagnosis based on critical systems-like behavior

This invention is a method, and system for predicting the onset of a seizure prior to electrograph onset in an individual. During an &ldquo;off-line&rdquo; mode, signals representing brain activity of an individual (either stored or real time) are collected, and features are extracted from those signals. A subset of features, which comprise a feature vector, are selected by a predetermined process to most efficiently predict (and detect) a seizure in that individual. An intelligent prediction subsystem is also trained &ldquo;off-line&rdquo; based on the feature vector derived from those signals. During &ldquo;on-line&rdquo; operation, features are continuously extracted from real time brain activity signals to form a feacture vector, and the feature vector is continuously analyzed with the intelligent prediction subsystem to predict seizure onset in a patient. The system, and method are preferably implemented in an implanted device ( ) that is capable of warning externally an individual of the probability of a seizure, and/or automatically taking preventative actions to abort the seizure. In addition, methods are provided for applying intervention measures to an animal to abort or modulat a seizure by adjusting the modality of an intervention measure; and/or parameters of an intervention measure based upon a probability measure indicative of a likelihood of seizure occurrence; and/or a predicted time to seizure onset.

Method and apparatus for predicting the onset of seizures based on features derived from signals indicative of brain activity

A method and apparatus is provided which analyzes an image of an object to detect and identify defects in the object utilizing multi-dimensional wavelet neural networks. &ldquo;The present invention generates a signal representing part of the object, then extracts certain features of the signal. These features are then provided to a multidimensional neural network for classification, which indicates if the features correlate with a predetermined pattern. This process of analyzing the features to detect and identify predetermined patterns results in a robust fault detection and identification system which is computationally efficient and economical because of the learning element contained therein which lessens the need for human assistance.&rdquo;

Method and apparatus for analyzing an image to detect and identify patterns

Growing community of inventors

Alpharetta, GA, United States of America

Javier Echauz