Blogs (1) >>
ASE 2019
Sun 10 - Fri 15 November 2019 San Diego, California, United States

Deep neural networks (DNNs) are shown to be promising solutions in many challenging artificial intelligence tasks, including object recognition, natural language processing, and even unmanned driving. A DNN model, generally based on statistical summarization of in-house training data, aims to predict correct output given an input encountered in the wild. In general, 100% precision is therefore impossible due to its probabilistic nature. For DNN practitioners, it is very hard, if not impossible, to figure out whether the low precision of a DNN model is an inevitable result, or caused by defects such as bad network design or improper training process. This paper aims at addressing this challenging problem. We approach with a careful categorization of the root causes of low precision. We find that the internal data flow footprints of a DNN model can provide insights to locate the root cause effectively. We then develop a tool, namely, DeepMorph (DNN Tomography) to analyze the root cause, which can instantly guide a DNN developer to improve the model. Case studies on four popular datasets show the effectiveness of DeepMorph.

Wed 13 Nov

Displayed time zone: Tijuana, Baja California change

15:20 - 16:00
Poster Session: Late Breaking ResultsLate Breaking Results at Kensington Ballroom
15:20
40m
Poster
Recommendation of Exception Handling Code in Mobile App Development
Late Breaking Results
Pre-print
15:20
40m
Poster
LVMapper: A Large-variance Clone Detector Using Sequencing Alignment Approach
Late Breaking Results
Ming Wu , Pengcheng Wang University of Science and Technology of China, Kangqi Yin , Haoyu Cheng , Yun Xu University of Science and Technology of China, Chanchal K. Roy University of Saskatchewan
Pre-print
15:20
40m
Poster
K-CONFIG: Using Failing Test Cases to Generate Test Cases in GCC Compilers
Late Breaking Results
Pre-print Media Attached
15:20
40m
Poster
An Empirical Study on the Characteristics of Question-Answering Process on Developer Forums
Late Breaking Results
Yi Li Nanyang Technological University, Shaohua Wang New Jersey Institute of Technology, USA, Tien N. Nguyen University of Texas at Dallas, Son Nguyen The University of Texas at Dallas, Xinyue Ye , Yan Wang
Pre-print
15:20
40m
Poster
Testing Neural Programs
Late Breaking Results
Md Rafiqul Islam Rabin University of Houston, Ke Wang Visa Research, Mohammad Amin Alipour
Pre-print Media Attached
15:20
40m
Poster
Self Learning from Large Scale Code Corpus to Infer Structure of Method Invocations
Late Breaking Results
Pre-print
15:20
40m
Poster
Data Sanity Check for Deep Learning Systems via Learnt Assertions
Late Breaking Results
Haochuan Lu Fudan University, Huanlin Xu , Nana Liu , Yangfan Zhou Fudan University, Xin Wang
Pre-print
15:20
40m
Poster
Software Engineering for Fairness: A Case Study with Hyperparameter Optimization
Late Breaking Results
Joymallya Chakraborty North Carolina State University, Tianpei Xia , Fahmid M. Fahid , Tim Menzies North Carolina State University
Pre-print
15:20
40m
Poster
API Misuse Correction: A Statistical Approach
Late Breaking Results
Pre-print
15:20
40m
Poster
Should We Add Repair Time to an Unfixed Bug? An Exploratory Study of Automated Program Repair on 2980 Small-Scale Programs
Late Breaking Results
Pre-print
15:20
40m
Poster
Learning test traces
Late Breaking Results
Pre-print
15:20
40m
Poster
The Dynamics of Software Composition Analysis
Late Breaking Results
Pre-print
15:20
40m
Poster
A Process Mining based Approach to Improving Defect Detection of SysML Models.
Late Breaking Results
Mounifah Alenazi , Nan Niu University of Cincinnati, Juha Savolainen Danfoss
Pre-print
15:20
40m
Poster
Open-Source Projects and their Collaborative Development Workflows
Late Breaking Results
panuchart bunyakiati kasetsart university, Usa Sammapun kasetsart university
Pre-print
15:20
40m
Poster
Detecting Deep Neural Network Defects with Data Flow Analysis
Late Breaking Results
Jiazhen Gu , Huanlin Xu , Yangfan Zhou Fudan University, Xin Wang , Hui Xu , Michael Lyu The Chinese University of Hong Kong
Pre-print
15:20
40m
Poster
On building an automated responding system for app reviews: What are the characteristics of reviews and their responses?
Late Breaking Results
Pre-print