Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Matrix Factorization Techniques for Recommender Systems

From the Book:
“Clean code that works” is Ron Jeffries’ pithy phrase. The goal is clean code that works, and for a whole bunch of reasons:
Clean code that works is a predictable way to develop. You know when you are finished, without having to worry about a long bug trail.Clean code that works gives you a chance to learn all the lessons that the code has to teach you. If you only ever slap together the first thing you think of, you never have time to think of a second, better, thing. Clean code that works improves the lives of users of our software.Clean code that works lets your teammates count on you, and you on them.Writing clean code that works feels good.But how do you get to clean code that works? Many forces drive you away from clean code, and even code that works. Without taking too much counsel of our fears, here’s what we do—drive development with automated tests, a style of development called “Test-Driven Development” (TDD for short). 
In Test-Driven Development, you:
Write new code only if you first have a failing automated test.Eliminate duplication.

Two simple rules, but they generate complex individual and group behavior. Some of the technical implications are:You must design organically, with running code providing feedback between decisionsYou must write your own tests, since you can’t wait twenty times a day for someone else to write a testYour development environment must provide rapid response to small changesYour designs must consist of many highly cohesive, loosely coupled components, just to make testing easy

The two rules imply an order to the tasks ofprogramming:
1. Red—write a little test that doesn’t work, perhaps doesn’t even compile at first
2. Green—make the test work quickly, committing whatever sins necessary in the process
3. Refactor—eliminate all the duplication created in just getting the test to work


Red/green/refactor. The TDD’s mantra.

Assuming for the moment that such a style is possible, it might be possible to dramatically reduce the defect density of code and make the subject of work crystal clear to all involved. If so, writing only code demanded by failing tests also has social implications:
If the defect density can be reduced enough, QA can shift from reactive to pro-active workIf the number of nasty surprises can be reduced enough, project managers can estimate accurately enough to involve real customers in daily developmentIf the topics of technical conversations can be made clear enough, programmers can work in minute-by-minute collaboration instead of daily or weekly collaborationAgain, if the defect density can be reduced enough, we can have shippable software with new functionality every day, leading to new business relationships with customers


So, the concept is simple, but what’s my motivation? Why would a programmer take on the additional work of writing automated tests? Why would a programmer work in tiny little steps when their mind is capable of great soaring swoops of design? Courage.
Courage
Test-driven development is a way of managing fear during programming. I don’t mean fear in a bad way, pow widdle prwogwammew needs a pacifiew, but fear in the legitimate, this-is-a-hard-problem-and-I-can’t-see-the-end-from-the-beginning sense. If pain is nature’s way of saying “Stop!”, fear is nature’s way of saying “Be careful.” Being careful is good, but fear has a host of other effects:
Makes you tentativeMakes you want to communicate lessMakes you shy from feedbackMakes you grumpy

None of these effects are helpful when programming, especially when programming something hard. So, how can you face a difficult situation and:
Instead of being tentative, begin learning concretely as quickly as possible.Instead of clamming up, communicate more clearly.Instead of avoiding feedback, search out helpful, concrete feedback.(You’ll have to work on grumpiness on your own.) 

Imagine programming as turning a crank to pull a bucket of water from a well. When the bucket is small, a free-spinning crank is fine. When the bucket is big and full of water, you’re going to get tired before the bucket is all the way up. You need a ratchet mechanism to enable you to rest between bouts of cranking. The heavier the bucket, the closer the teeth need to be on the ratchet.

The tests in test-driven development are the teeth of the ratchet. Once you get one test working, you know it is working, now and forever. You are one step closer to having everything working than you were when the test was broken. Now get the next one working, and the next, and the next. By analogy, the tougher the programming problem, the less ground should be covered by each test.

Readers of Extreme Programming Explained will notice a difference in tone between XP and TDD. TDD isn’t an absolute like Extreme Programming. XP says, “Here are things you must be able to do to be prepared to evolve further.” TDD is a little fuzzier. TDD is an awareness of the gap between decision and feedback during programming, and techniques to control that gap. “What if I do a paper design for a week, then test-drive the code? Is that TDD?” Sure, it’s TDD. You were aware of the gap between decision and feedback and you controlled the gap deliberately.

That said, most people who learn TDD find their programming practice changed for good. “Test Infected” is the phrase Erich Gamma coined to describe this shift. You might find yourself writing more tests earlier, and working in smaller steps than you ever dreamed would be sensible. On the other hand, some programmers learn TDD and go back to their earlier practices, reserving TDD for special occasions when ordinary programming isn’t making progress.

There are certainly programming tasks that can’t be driven solely by tests (or at least, not yet). Security software and concurrency, for example, are two topics where TDD is not sufficient to mechanically demonstrate that the goals of the software have been met. Security relies on essentially defect-free code, true, but also on human judgement about the methods used to secure the software. Subtle concurrency problems can’t be reliably duplicated by running the code.

Once you are finished reading this book, you should be ready to:
Start simplyWrite automated testsRefactor to add design decisions one at a time

This book is organized into three sections.
An example of writing typical model code using TDD. The example is one I got from Ward Cunningham years ago, and have used many times since, multi-currency arithmetic. In it you will learn to write tests before code and grow a design organically.An example of testing more complicated logic, including reflection and exceptions, by developing a framework for automated testing. This example also serves to introduce you to the xUnit architecture that is at the heart of many programmer-oriented testing tools. In the second example you will learn to work in even smaller steps than in the first example, including the kind of self-referential hooha beloved of computer scientists.Patterns for TDD. Included are patterns for the deciding what tests to write, how to write tests using xUnit, and a greatest hits selection of the design patterns and refactorings used in the examples.

I wrote the examples imagining a pair programming session. If you like looking at the map before wandering around, you may want to go straight to the patterns in Section 3 and use the examples as illustrations. If you prefer just wandering around and then looking at the map to see where you’ve been, try reading the examples through and refering to the patterns when you want more detail about a technique, then using the patterns as a reference.

Several reviewers have commented they got the most out of the examples when they started up a programming environment and entered the code and ran the tests as they read.

A note about the examples. Both examples, multi-currency calculation and a testing framework, appear simple. There are (and I have seen) complicated, ugly, messy ways of solving the same problems. I could have chosen one of those complicated, ugly, messy solutions to give the book an air of “reality.” However, my goal, and I hope your goal, is to write clean code that works. Before teeing off on the examples as being too simple, spend 15 seconds imagining a programming world in which all code was this clear and direct, where there were no complicated solutions, only apparently complicated problems begging for careful thought. TDD is a practice that can help you lead yourself to exactly that careful thought.

Test Driven Development: By Example

Genetic programming (GP) is a systematic, domain-independent method for getting computers to solve problems automatically starting from a high-level statement of what needs to be done. Using ideas from natural evolution, GP starts from an ooze of random computer programs, and progressively refines them through processes of mutation and sexual recombination, until high-fitness solutions emerge. All this without the user having to know or specify the form or structure of solutions in advance. GP has generated a plethora of human-competitive results and applications, including novel scientific discoveries and patentable inventions. This unique overview of this exciting technique is written by three of the most active scientists in GP. See www.gp-field-guide.org.uk for more information on the book.

/pdf/a-field-guide-to-genetic-programming-3wkkmsw81n.pdf

A Field Guide to Genetic Programming

Rough sets theory for multicriteria decision analysis

The condition of the vascular network of human eye is an important diagnostic factor in ophthalmology. Its segmentation in fundus imaging is a nontrivial task due to variable size of vessels, relatively low contrast, and potential presence of pathologies like microaneurysms and hemorrhages. Many algorithms, both unsupervised and supervised, have been proposed for this purpose in the past. We propose a supervised segmentation technique that uses a deep neural network trained on a large (up to 400 $\thinspace$ 000) sample of examples preprocessed with global contrast normalization, zero-phase whitening, and augmented using geometric transformations and gamma corrections. Several variants of the method are considered, including structured prediction, where a network classifies multiple pixels simultaneously. When applied to standard benchmarks of fundus imaging, the DRIVE, STARE, and CHASE databases, the networks significantly outperform the previous algorithms on the area under ROC curve measure (up to $ > {0.99}$ ) and accuracy of classification (up to $ > {0.97}$ ). The method is also resistant to the phenomenon of central vessel reflex, sensitive in detection of fine vessels ( ${\rm sensitivity} > {0.87}$ ), and fares well on pathological cases.

Segmenting Retinal Blood Vessels With Deep Neural Networks

Traditional Genetic Programming (GP) searches the space of functions/programs by using search operators that manipulate their syntactic representation, regardless of their actual semantics/behaviour. Recently, semantically aware search operators have been shown to outperform purely syntactic operators. In this work, using a formal geometric view on search operators and representations, we bring the semantic approach to its extreme consequences and introduce a novel form of GP --- Geometric Semantic GP (GSGP) --- that searches directly the space of the underlying semantics of the programs. This perspective provides new insights on the relation between program syntax and semantics, search operators and fitness landscape, and allows for principled formal design of semantic search operators for different classes of problems. We derive specific forms of GSGP for a number of classic GP domains and experimentally demonstrate their superiority to conventional operators.

/pdf/geometric-semantic-genetic-programming-5gdn85uq8d.pdf

Geometric semantic genetic programming

Genetic programming (GP) is not a field noted for the rigor of its benchmarking. Some of its benchmark problems are popular purely through historical contingency, and they can be criticized as too easy or as providing misleading information concerning real-world performance, but they persist largely because of inertia and the lack of good alternatives. Even where the problems themselves are impeccable, comparisons between studies are made more difficult by the lack of standardization. We argue that the definition of standard benchmarks is an essential step in the maturation of the field. We make several contributions towards this goal. We motivate the development of a benchmark suite and define its goals; we survey existing practice; we enumerate many candidate benchmarks; we report progress on reference implementations; and we set out a concrete plan for gathering feedback from the GP community that would, if adopted, lead to a standard set of benchmarks.

/pdf/genetic-programming-needs-better-benchmarks-2yjk1wqckx.pdf

Genetic programming needs better benchmarks

This paper presents an empirical study of the use of the rough set approach to reduction of data for a neural network classifying objects described by quantitative and qualitative attributes. Two kinds of reduction are considered: reduction of the set of attributes and reduction of the domains of attributes. Computational tests were performed with five data sets having different character, for original and two reduced representations of data. The learning time acceleration due to data reduction is up to 4.72 times. The resulting increase of misclassification error does not exceed 11.06%. These promising results let us claim that the rough set approach is a useful tool for preprocessing of data for neural networks.

Rough set reduction of attributes and their domains for neural networks

In this paper we use genetic programming for changing the representation of the input data for machine learners. In particular, the topic of interest here is feature construction in the learning-from-examples paradigm, where new features are built based on the original set of attributes. The paper first introduces the general framework for GP-based feature construction. Then, an extended approach is proposed where the useful components of representation (features) are preserved during an evolutionary run, as opposed to the standard approach where valuable features are often lost during search. Finally, we present and discuss the results of an extensive computational experiment carried out on several reference data sets. The outcomes show that classifiers induced using the representation enriched by the GP-constructed features provide better accuracy of classification on the test set. In particular, the extended approach proposed in the paper proved to be able to outperform the standard approach on some benchmark problems on a statistically significant level.

Krzysztof Krawiec

Papers

Segmenting Retinal Blood Vessels With Deep Neural Networks

Geometric semantic genetic programming

Genetic programming needs better benchmarks

Rough set reduction of attributes and their domains for neural networks

Genetic Programming-based Construction of Features for Machine Learning and Knowledge Discovery Tasks