Recent Publications
Under Review – NeurIPS 2020 An Adversarial Approach for Explaining the Predictions of Deep Neural Networks
Abstract: Machine learning models have been successfully applied to a wide range of applications including computer vision, natural language processing, and speech recognition. A successful implementation of these models however, usually relies on deep neural
CVPR 2020 Robust Design of Deep Neural Networks Against Adversarial Attacks Based on Lyapunov Theory
Abstract: Deep neural networks (DNNs) are vulnerable to subtle adversarial perturbations applied to the input. These adversarial perturbations, though imperceptible, can easily mislead the DNN. In this work, we take a control theoretic approach to
KDD 2019 Connecting Lyapunov Control Theory to Adversarial Attacks
Abstract: Significant work is being done to develop the math and tools necessary to build provable defenses, or at least bounds, against adversarial attacks of neural networks. In this work, we argue that tools from
Modzy Labs Innovations
Experience the innovations that started in the Lab and often become unique features in the Modzy Enterprise AI Platform. These innovations help accelerate value creation for our customers and advance the state of Artificial Intelligence for everyone. Schedule a Demo to see these features in action!
Explainability at Modzy
Modzy’s patent-pending approach to explainability is the fastest in the world. Listen to Modzy’s Head of Science, Clayton Davis, discuss how explainability factors into our approach to building trustworthy...
Adversarial Defense
Modzy has patent-pending adversarial defense built in to ensure that your models and data remain secure . Listen to Modzy’s Head of Applied AI Research, Dr. Arash Rahnama, discuss...
Explainable AI
Model interpretability is the degree to which a human can understand the cause of a decision made by the model [1]. When a model makes a decision, there must...
Adversarial AI
Deep learning is the main force behind the recent advances in the field of artificial intelligence (AI). Deep learning models are capable of performing on par with, if not...
Retraining With Your Own Data
In simple terms, transfer learning is a machine learning approach where a model that is already trained on a specific data set and developed for a specific task is...
Training Data Security
The effectiveness and predictive power of machine learning models is highly dependent on the quality of data used during the training phase. In most real-world scenarios, models are trained...
Data Imbalance
Imbalanced classes are a common problem in machine learning classification where the training dataset contains a disproportionate ratio of samples in each class. Examples of real-world scenarios that suffer...
Training Robust AI Models – A...
We’ve developed a new training method to ensure that adversarial defense is built into AI models by ensure models are robust and resilient. Listen to Modzy’s Head of Applied...
Data Privacy: Deployment
Protecting data privacy is one of the most important issues in technology today. From data breaches, to questions about ethical data sharing, it is clear that access to data...
Meta Learning
A well-performing machine learning model often requires training on a very large data set of labeled data points. Humans, in contrast, learn new concepts faster and more efficiently from...
Human In The Loop
AI models are being deployed every day to support the human decision-making process in high-stakes domains. In many environments, an AI model is the sole decision-maker or a user’s...
Models At The Edge
Edge computing refers to analyzing and processing data near where the data is generated, to decrease data flow and thereby reduce network traffic and response time. With the recent...
