Reinforcement Learning To Write Java Unit Tests Mathew Lodge

Diffblue: AI For Code 1

Talk in a nutshell: In the same way AI can search for the best Go move, AI can search for the best unit test

Continuous Integration In Theory Write Code Commit Unit Tests Integrate Full Tests Deploy

Continuous Integration in Practice Uncaught Errors Result in Broken Pipelines and Low Velocity Write Code Commit Tests? Integrate Full Tests Time to find & fix errors Minutes Hours Hours to Days

What is AI?

Code Transforms Input to Output Q: 26,464*36,361/4? A: 240,564,376

Limitations Of Human Coding Q: What’s this? A: ????

AI: Computer Iterates on Statistical Model That Relates Input To Output AI Model A: Q: What’s this? 91% Cat 23% Tabby 86% Dog 32% Spaniel 11% Pebble

AI: Not Artificial, Not Intelligent • Not Artificial: Solves real problems programmers cannot reach • Not Intelligent: Just maths • Great when you can’t brute-force the problem

Reinforcement Learning AI in AlphaGo

Go Is A Hard Game To Automate • Far more possible moves than Chess: ~10 170 • ~10 92 atoms in the known universe • Long-term strategy makes a big difference in Go • Hard to tell if a move is good until much later

Reinforcement Learning In General Observation Action Environment Reward Update Evaluate Model State update

AlphaGo Go Moves Algorithm Play Game Reward Updated AlphaGo Neural Evaluation Networks State update

AlphaGo Is Probabilistic Search Of The Go Gameplay Space • Monte Carlo to generate potential moves • “Policy” neural network predicts best next move • “Value” neural network predicts who will win

Monte Carlo With Prediction

What Does This Have To Do With Java Testing Or DevOps?

Testing As Search Of Program Space Set of All Possible Test Programs Programs that are valid and run High coverage tests Tests that developers find easy to read

Test Writing Is Hard To Automate • Exponential search space • Complex interdependencies between program and data • Hard to tell if a test is “good” • Practical difficulties: IO, frameworks, dynamic typing, semantic understanding, idiomatic code

Diffblue Reinforcement Learning Coverage + Results Test Candidate JVM Reward Existing Java code Test mutation Evaluator Test Writer

Unsupervised Learning Very Fast Can Run On A Dev Laptop

Diffblue AI Writes Unit Test Suites • Run quickly • Run early • Find unit-level errors • Improve coverage • Are easy to understand

Typical Software Lifecycle Engineer No updates Engineer Is the code Running Pull Request Run tests writes code change code change correct? PR approved Yes

Lifecycle with Diffblue Cover Engineer No updates Engineer Is the code Running Run tests + Pull Request writes code change code Diffblue suite change correct? PR approved Yes Regression Unit AI writes Diffblue AI Test Suite tests from updates Unit current code Test Suite

Diffblue At Goldman Sachs . 2x better test coverage (70%) . 10x faster than manual coding

Summary • Lack of tests to run early in pipeline is a key DevOps blocker • Diffblue Cover AI writes unit tests for your current Java code • Cover is 10-100x faster than manual test writing

Free trials available at Diffblue.com Any Questions?

Reinforcement Learning To Write Java Unit Tests Mathew Lodge - PowerPoint PPT Presentation

Reinforcement Learning To Write Java Unit Tests Mathew Lodge Diffblue: AI For Code 1 Talk in a nutshell: In the same way AI can search for the best Go move, AI can search for the best unit test Continuous Integration In Theory Write Code

Unit Testing Why and How to Write Unit Tests in KDE? David Faure faure@kde.org Kvin Ottens

Unit Tests Are Not Enough Practical Functional Testing in Java Martin Klose klose brothers

or how to write tests so that they serve you well whoami ~16 years as a developer, ~12 years

What Are Unit Tests and Why Write Them? Aime Gott Education Practice Lead, Mango Solutions

RL Overview of topics About Reinforcement Learning The Reinforcement Learning Problem

JAVA Basics & OOP 2020/3/14 Write Once, Run Anywhere Java Virtual Machine (JVM)

Jartege : a Tool for Random Generation of Unit Tests for Java Classes Catherine Oriat LSR/IMAG,

Reinforcement Learning Reinforcement Learning Reinforcement Learning in a nutshell g Imagine

Reinforcement Learning You can think of supervised learning as the teacher providing answers

Examples of Reinforcement Learning Robocup Soccer Teams Stone & Veloso, Reidmiller et al.

CS885 Reinforcement Learning Module 2: June 6, 2020 Maximum Entropy Reinforcement Learning

Unit Tests: Using PHPUnit to Test Your Code With Your Host Juan Treminio http://jtreminio.com

Reinforcement Learning Reinforcement Learning Now that you know a little about Optimal Control

Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning II Still

R i f R i f Reinforcement Learning III Reinforcement Learning III t L t L i i III III Dec

Reinforcement Learning UMaine COS 470/570 Introduction to AI Why reinforcement learning?

7. Motor Control and Reinforcement Learning Outline A. Action Selection and Reinforcement B.

UNIT TESTING 3 / 8 1 / 8 Unit testing involves: Lots of small, independent tests Reporting

Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement Learning and Tree

Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning Ja Jan-Wi

Reinforcement Learning AIMA Chapters: 21.1, 21.2, 21.3. Sutton and Barto, Reinforcement Learning:

Module 11 Introduction to Reinforcement Learning CS 886 Sequential Decision Making and

1 Pure Reinforcement Learning vs. Reinforcement Learning Monte-Carlo Planning No knowledge

Safe Reinforcement Learning Philip S. Thomas Stanford CS234: Reinforcement Learning, Guest