Analyzing Software using Deep Learning - Software Lab - Department of Computer Science

Analyzing Software using Deep Learning (Summer Semester 2020)

Quick Facts

Lecturer	Prof. Dr. Michael Pradel
Teaching assistants	Jibesh Patra
Course type	Lecture + Exercises
Language	English
Ilias	Ilias course with forum

Content

Software developers use tools that automate particular subtasks of the development process. Recent advances in machine learning, in particular deep learning, are enabling tools that had seemed impossible only a few years ago, such as tools that predict what code to write next, which parts of a program are likely to be incorrect, and how to fix software bugs. This course introduces recent techniques developed at the intersection of program analysis and machine learning. In one part of the course, we will cover some basics of both fields, followed by a discussion of several recent deep learning-based programming tools. In the other part of the course, students will implement their own deep learning-based program analysis based on an existing framework. Grading will be based on the implementation as well as a written exam.

Online Course

The course as an online course. All lecture videos will be provided in this playlist throughout the semester, roughly following the schedule below.

Schedule

This is a preliminary schedule and may be subject to change. There are no exercise sessions, but students will work on a course project under supervision by a teaching assistant.

Date	Topic	Material	Deadlines and special events
Apr 23, 2020	Introduction	Part 1 (Motivation): Lecture, Slides Part 2 (Organization): Lecture, Slides Part 3 (Basics): Lecture, Slides Tutorial: Deep Learning with PyTorch
Apr 30, 2020	RNN-based code completion and repair	Part 1 (RNNs): Lecture, Slides Part 2 (Code Completion): Lecture, Slides, Paper Part 3 (Automated Repair): Lecture, Slides, Paper Tutorial: RNNs in PyTorch
May 7, 2020	Reasoning about types and code changes with hierarchical neural networks	Part 1 (Hierarchical Models): Lecture, Slides Part 2 (Type Prediction): Lecture, Slides, Paper Part 3 (Code Changes): Lecture, Slides, Paper Tutorial: Esprima JavaScript Parser
May 14, 2020	Sequence-to-sequence networks and their applications	Part 1 (Sequence-to-sequence Models): Lecture, Slides Part 2 (API Usages): Lecture, Slides, Paper Part 3 (Interpreting Python): Lecture, Slides, Paper Tutorial: Sequence-to-sequence in PyTorch
May 22, 2020	Start of course project
June 11, 2020	Summarizing programs with convolutional networks	Part 1 (Convolutional Neural Networks): Lecture, Slides Part 2 (AST Classification): Lecture, Slides, Paper
June 25, 2020	Token vocabulary and code embeddings	Part 1 (Vocabulary problem): Lecture, Slides Part 2 (Pre-trained embeddings): Lecture, Slides Part 3 (PL/NL): Lecture, Slides Papers: Mikolov et al., Karampatsis et al., Gu et al.
July 2, 2020	AST-based and graph-based neural networks of code	Part 1 (Introduction): Lecture, Slides Part 2 (AST paths): Talk by Eran Yahav, Paper Part 3 (Graph neural networks): Talk by Miltos Allamanis, Paper
July 17, 2020			Deadline for course project
July 20-24, 2020	Final presentations of course projects
TBD	--		Exam

Course Project

The course project is about implementing a learning-based bug detector. Please see the project description for details.