Virtual Seminar: Experimenting with Classifiers
Speaker: Martin Shepperd
Overview:
1. Much machine learning research is empirical in nature (analytic solutions are intractable).
2. So we conduct experiments…
3. where the competing algorithms are *treatments*, the datasets are *experimental units* and classification performance is the *response measure*,
4. and the experimental design is typically *repeated measures*.
5. This view of machine learning research should inform the study design and analysis.
6. Failure to do so helps explain the unreliability of many published `results’.
Virtual Seminar: Static Microservice Architecture Recovery Using Model-Driven Engineering
Speaker: Nuha Alshuqayran
Abstract: In recent years, the software development industry has witnessed effective changes which have led to the development of new architectural styles of software. In this respect, this seminar presents research work aimed to support the microservice architectural style. Software developed using the microservice architecture is complex and distributed and involves several technologies and components. Reverse engineering and specifically Architecture Recovery can aid in the understanding and maintenance of microservice systems. We present how we are developing our MicroService Architecture Recovery (MiSAR) approach, which allows software engineers to recover architectures of microservice systems. MiSAR follows Model Driven Engineering and includes different models such as modelling languages for Microservice architecture and undertakes transformation of models as a set of mapping rules for microservice based systems. In the seminar, we will demonstrate how our approach is capable of obtaining expressive architectural models of microservice systems in an effective way.
Virtual Seminar: Multiparty Session Programming with Global Protocol Combinators
Speaker: Dr Rumyana Neykova
Abstract: Checking compatibility of concurrent programs, i.e. if two or more processes can communicate without errors, is a pressing problem in the verification community. State-of-the-art verification tools are limited to model-checkers and SMT solvers, which are foreign to many developers and too computationally expensive to use in practice.
In this talk, I will demonstrate a new approach to programming and verification of concurrent and distributed programs. The approach relies on a few compositional constructs, called global protocol combinator (GPC). I will show an encoding of a popular theory for communicating programs into GPC. Such encoding reveals that the problem of checking compatibility can be reduced to the standard problem of variant/record subtyping. This realisation allows type systems of existing general-purpose programming languages to be utilised for static detection of concurrency bugs, without the need for external model checkers. I will show an implementation of our encoding in native OCaml and will discuss its expressive power and performance. We have tested the approach by implementing and verifying a plethora of concurrency algorithms, as well as several popular communication protocols (DNS, OAuth, and SMPT).
We are recruiting two new funded PhD Students.
Project One
Securing Microservices with Just-In-Time Model Verification – Dr Nour Ali & Dr Rumyana Neykova – Project details here
Project Two
The REWIND Project (Does REfactoring Software Work? An INDustrial and Open-source Approach) – Professor Steve Counsell & Dr Mahir Arzoky – Project details here
Application Guidelines Please click here to download.
Application Deadline 29/05/2020
IDA meeting held at WLFB 207/208 (2nd floor of Wilfred Brown) at 3:00PM
Talks from:
Leila Yousefi: The Prevalence of Errors in Machine Learning Experiments (slides can be found here)
Marco Ortu: The Butterfly “Affect”: Impact of Development Practices on Cryptocurrency Prices (slides can be found here)
Gabriel Scali: Constraint Satisfaction Problems and Constraint Programming (slides can be found here)
