Write a Blog >>
Wed 11 Nov 2020 17:35 - 17:36 at Virtual room 2 - ML Model Building

Deep neural networks (DNNs) have been widely applied in the software development process to automatically learn patterns from massive data. However, many applications still make decisions based on rules that are manually crafted and verified by domain experts due to safety or security concerns.
In this paper, we aim to close the gap between DNNs and rule-based systems by automating the rule generation process via extracting knowledge from well-trained DNNs. Existing techniques with similar purposes either rely on specific DNNs input instances or use inherently unstable random sampling of the input space. Therefore, these approaches either limit the exploration area to a local decision-space of the DNNs or fail to converge to a consistent set of rules. The resulting rules thus lack representativeness and stability.

In this paper, we address the two aforementioned shortcomings by discovering a global property of the DNNs and use it to remodel the DNNs decision-boundary. We name this property as the activation probability, and show that this property is stable.
With this insight, we propose an approach named DENAS including a novel rule-generation algorithm. Our proposed algorithm approximates the non-linear decision boundary of DNNs by iteratively superimposing a linearized optimization function.

We evaluate the representativeness, stability, and accuracy of DENAS against five state-of-the-art techniques (LEMNA, Gradient, IG, DeepTaylor, and DTExtract) on three software engineering and security applications: Binary analysis, PDF malware detection, and Android malware detection. Our results show that DENAS can generate more representative rules consistently in a more stable manner over other approaches. We further offer case studies that demonstrate the applications of DENAS such as debugging faults in the DNNs and generating signatures that can detect zero-day malware.

Wed 11 Nov

Displayed time zone: (UTC) Coordinated Universal Time change

17:30 - 18:00
17:30
2m
Talk
AMS: Generating AutoML Search Spaces from Weak Specifications
Research Papers
Jose Cambronero Massachusetts Institute of Technology, USA, Jürgen Cito TU Wien and MIT, Martin C. Rinard Massachusetts Institute of Technology, USA
DOI
17:33
1m
Talk
Continuous Experimentation on Artificial Intelligence Software: A Research Agenda
Visions and Reflections
Anh Nguyen-Duc University of South Eastern Norway, Pekka Abrahamsson University of Jyväskylä
DOI
17:35
1m
Talk
DENAS: Automated Rule Generation by Knowledge Extraction from Neural Networks
Research Papers
SiminChen University of Texas at Dallas, USA, Soroush Bateni University of Texas at Dallas, USA, Sampath Grandhi University of Texas at Dallas, USA, Xiaodi Li University of Texas at Dallas, USA, Cong Liu University of Texas at Dallas, USA, Wei Yang University of Texas at Dallas, USA
DOI
17:37
1m
Talk
On Decomposing a Deep Neural Network into ModulesACM SIGSOFT Distinguished Paper Award
Research Papers
Rangeet Pan Iowa State University, USA, Hridesh Rajan Iowa State University, USA
DOI Media Attached
17:39
1m
Talk
Synthesizing Correct Code for Machine Learning Programs
Student Research Competition
Joshua Gisi North Dakota State University, USA
DOI
17:41
19m
Talk
Conversations on ML Model Building
Paper Presentations
Jose Cambronero Massachusetts Institute of Technology, USA, Rangeet Pan Iowa State University, USA, Simin Chen , Wei Yang University of Texas at Dallas, USA, M: John-Paul Ore North Carolina State University