With the development and growth of the Institute of Technologies and Software Development and its activities, seminars become more popular and expand its auditorium. That’s why the new name for seminars was created and we are glad to welcome you at ACSIS – Advanced Computer Science Innopolis Seminars. 

ACSIS is a bi-weekly meeting organized by Innopolis University's Institutes with the idea of keeping informing each other of what we are doing and help each other to improve. The range of topics offered by ACSIS is quite vast since it involves all Innopolis' Institutes: it offers topics from Software Engineering, Industrial Production, Information Systems, etc.

It is intended to students, researchers, engineers, industrial experts.

Practical details

Unless indicated otherwise, sessions take place at Innopolis University, 3rd floor, room 307 at 5.10 p.m.

If you have any questions/comments you can contact Victor Rivera at v.rivera@innopolis.ru 


ACSIS schedule for 2016 - 2017 

22/03/2017 - Gabriele Trovato 

05/04/2017 - Luca Longo 


Upcoming session

22/03/2017- Gabriele Trovato

Title: "Procedurally Generated History with Sid Meier's Civilization"

Abstract

Generating interesting game worlds is an ongoing challenge within Procedural Content Generation. While the use of algorithms for automatically generating content could potentially bring many advantages, the content itself is often random and not well-crafted. In this presentation, a potential solution to this issue, implemented in a modpack for Sid Meier’s Civilization IV, is explored by creating meaning through procedurally tracing a history of the game.  


Past session 

23/11/2016 - Mohamad Kassab

Title: "The Changing Landscape of Requirements Engineering Practices over the Past Decade"

Abstract:

Even though there is ample information available on solid requirements engineering practices, anecdotal evidence still indicates poor practices in industry. The key issue in implementing an improvement is to first identify the areas that need most improvement. Three surveys were conducted in the past decade on the state of practice of requirements engineering. Surveys data obtained includes characteristics of projects, practices, organizations, and practitioners related to requirements engineering.  In this talk, a comparison and analysis of the responses from the three surveys will be discussed in order to understand the changing landscape of requirements engineering industrial practices over the past years.



09/11/2016 - Leonard Johard

Title: “Algorithmic mysteries in the human brain”

Abstract:

“Each second our brains solve a number of now well-established statistical problems. Modern machine learning techniques have been able to solve the very same problems with impressive practical results, often matching the abilities of human learners. Unfortunately, here the similarities end. Despite the neurons’ apparent simplicity, we have in these cases been unable to discover a single algorithm that works under even weak biological constraints. Leonard Johard will present open algorithmic challenges in the brain and how finding solutions will impact the future of AI and neuroscience.”


26/10/2016 - Nikolay Shilov

Title: BDI-agents in a Cloud of Services 

Abstract:

Multiagent algorithm is a knowledge-based distributed algorithm that solves some problem by means of cooperative work of agents. But in an individual agent’s perspective a multigent algorithm is a reactive and proactive knowledge/believe-based rational algorithm to achieve agent’s own desires. In the talk we will discuss couple of multiagent algorithms from epistemic, efficiency (performance) and scaleability perspectives. In particular we will discuss a multiagent algorithm for Discrete Resource Allocation Problem.

Speaker would like to conclude this theoretical talk by some words in memory of  one of pillars of Computer Science (http://www.springer.com/us/book/9783540781264​) Boris Trakhtenbrot (http://cacm.acm.org/news/207650-in-memoriam-boris-trakhtenbrot-1921-2016/fulltext) who passed away September 19, 2016.

BIO of speaker: 

Nikolay Shilov graduated Novosibirsk State University in 1983 and then got Ph.D. in Computer Science from Novosibirsk Computing Center in 1987. (By the way, Novosibirsk Computing Center is the only institution in the former SU where Turing Lauriate was employed;-) Nikolay had faculty and research positions in Russia and abroad (Australia, Kazakhstan, New Zealand, South Korea). Since university years Nikolay studies Theory of Programming and Applied Logic (in line with tradition founded in Novosibirsk by Boris Trakhtenbrot). 

12/10/2016 - Alexandr Naumchev

Title: Hybrid Requirements

Abstract:

“Popular notations for requirement specifications frequently target specific categories of stakeholders, specific software development models, or specific

development activities; the results can give out-of-sync or downright incompatible views. Hybrid Requirements, a new approach to specifying functional

requirements, decrease synchronization overhead, improve software quality, and increase software process agility. A development case illustrates these benefits,

and a discussion compares hybrid requirements to other approaches.”

28/09/2016 - Hamna Aslam 

Title: Towards a better gaming experience

Abstract: 

The experience of buying and playing a game should be worth the cost of the game. However, beautiful packaging merely leads to disappointment when a person finds confusing objects inside with a big rule book.
This is the motivation behind examining intuitive game design. The AI and Games Development Lab’s research focuses on finding and investigating factors that can lead to intuitive design. We define intuitive game design as being:“A game design that helps a player in exploring the game play process through objects and features of the game and does not require a large explanation in rule books”.
This talk examines the results of a set of studies where board games were presented with their rule books removed to the participants, from different parts of the world, gathered for the selection interviews in Innopolis University. The results from the participants gives an insight into the factors, e.g shape and color of objects, text on cards etc., that can lead to an intuitive game design and a better gaming experience.​

About speaker:

Hamna Aslam was born in Multan, Pakistan, on January 7, 1987. She received the B.Sc. (Hons. with first-class standing) in Computer Engineering from Bahauddin Zakariya University, Multan and the M.Sc. in Computer Engineering from the University of Engineering and Technology, Lahore, Pakistan in 2015. Previously she was a visiting lecturer at the Institute of Business Management, University of Engineering and Technology Lahore (UET), Pakistan. Besides, Hamna Aslam worked in the energy sector of the government of Pakistan in capacity of Software Developer. In 2015 she co-organized the China Pakistan Economic Corridor Seminars in Lahore, Pakistan. She is currently a researcher in the Artificial Intelligence in Game Development Lab, Innopolis University, Republic of Tatarstan, Russian Federation.


Past sessions 2015-2016 


All Kazan Software Engineering Seminars – are the series of seminars organized by the Innopolis University Institute of Technologies and Sotfware Development . It covers a wide range of software engineering topics, both practical and theoretical, talks are given both by local researchers and by visitors.

Most talks are recorded and may be accessed at the link below shortly after the offline delivery.

26.04.2016 - Salvatore Distefano

Title: Volunteering, Crowdsourcing and Citizen Science: an overview from the computing perspective

Abstract:


The digital revolution alongside with the evolution of social networking and portable platform has opened a lot of possibilities. In particular, the terms crowdsourcing and citizen science have acquired special meanings, referring to a new approach of performing tedious and repetitive work which was formerly being conducted by traditional dedicated employees. In this talk, such new trends are discussed by analyzing several experimental researches and practical examples. The key aspects and the multifaceted representations of the crowdsourcing phenomenon starting from definitions and underlying working principles and motivations will be specified. Some practical examples are then discussed focusing on all the most relevant forms of adoption of these approaches, such as volunteering contribution in online communities, indirect crowdsourcing and gamification.    


12.04.2016 - Leonard Johard

Title: BioDynaMo

Abstract:

Neuroscientists worldwide are turning to simulations for help in understanding the human brain. The EU, USA and Japan are all trying to unite their scientists behind huge efforts to map the machinery of the adult brain in ever finer detail. These static models of the adult brain follow the traditional reductionist approach and ignore the greatest promise of simulators: study of emergent behaviour.

We are developing a new simulator for the human cortex. We will capture the brain development all the way from a single cell into an adult cortex, with the structures being emergent rather than predefined. New models of plasticity self-organizes the sensory information pathways and allows effective learning of motor control. All this can be done in the cloud with no need for the cumbersome and expensive scientific grids common today.


29.03.2016 - Vladimir Ivanov (LIPS, RA) 

Title: Modeling the reliability of software developed in multiple releases

Abstract:

The reliability of software is studied for decades due to its high importance for the industry of software engineering. Usually, the reliability of software is measured as a number of defects or failures in a released software product or service. Many software reliability growth models (SRGMs) have been developed for a case of a single release. We propose a model able to describe the process of multiple release development and testing. We also provide a preliminary experimental results of the evaluation of the model on several datasets.


15.03.2016 - Mohamed Elwakil

Deadlocks Prediction

Abstract:
A deadlock is a very serious condition that may arise in concurrent programs.  A set of processes is deadlocked if each process in the set is waiting for an event that only another process in the set can cause. In this talk we will present two research tools that uses runtime analysis to predict deadlocks.


02.02.20160 Mohamed Kassab

An Empirical Study on the Requirements Engineering Practices for Agile Software Development
 
Abstract:
Collecting, understanding, and managing requirements are critical aspects in all development methods including agile methods as well. Nevertheless, little contemporary data exists for document actual practices of software professionals for software requirements engineering activities in agile environments. To remedy this deficiency and provide useful data to other researchers we conducted a survey study that drew participants from wide range of professions, industries and geographic locations. We filtered the survey responses according to the Software Development Life Cycle; then we analyzed how agile processes (compared to traditional waterfall model) deal with a paradox of requirements engineering. In this paper, we present this exploratory survey and its quantitative results.

19.01.2016 - Alexandr Chichigin

DSL-based binding of requirements, code and verification. Mbeddr example.

Abstract: 
On a sample project we will see how Mbeddr IDE implements requirements traceability, textual and graphical representations and formal verification on general ground of language-oriented development and extensible domain-specific languages.

26.11.2015 - Antonio Puliafito
Combining IoT and Cyber world: Tools and methodologies to manage a urban environment 

Abstract:
New services are rapidly being developed to interact with the surrounding environment, collecting data and applying management strategies. By adding smart objects into the mix, new opportunities arise for contextualization and geo-awareness. In this view, computing, storage, networking, and sensing become complementary aspects to be coordinated in order to provide new types of Sensors and Actuators as a Service (SAaaS), in a software-defined autonomic infrastructure fashion. The reference scenario is highly dynamic, since it can involve both static (sensor networks, RFID, etc.) and mobile (smartphone, PDA, etc.) heterogeneous objects and nodes that can randomly join and leave the connected environment in a volunteer contributed scenario. Smart objects have to be abstracted and virtualized in order to be provided under the guise of customisable virtual devices and corresponding services. Enabling technologies towards the envisaged goals are: abstraction and virtualization of objects; volunteer techniques for enrolment and distributed coordination; Cloud-like, service-oriented interfaces, autonomic techniques and semantic enablement.  

Given such background, this talk will address the problem of integrating the cyber and the physical world through a comprehensive platform implementing tools for smart devices integration, configuration and management in a Cloud-like fashion, as well as empowering development of novel services. 
Following a quick introduction on the MDSLab research group, with on-going projects and research directions, the talk will introduce Cloud computing principles, will present mechanisms to interact with the Internet of things and will provide examples of applications to Smart cities and related services.

12.11.2015 - Alexandr Naumchev
Refinement in Object-Oriented Programming

Abstract:

The notion of refinement has its origins in the field of state machines. The concept is developed in some techniques including Z and Event-B. Yet, these techniques are not ideal. The talk illustrates by an example how to perform refinement in a contracted object-oriented language. The results show that a programming language may be more suitable for expressing refinement than the classical modeling instruments.


29.10.2015 - Victor Rivera 

The Event-B Method: an introduction

Abstract:
There exist different methodologies for developing systems. The number of methodologies has increased over the time thanks to the need of building correct software (specially critical systems). During this talk, I will be presenting a different approach to model systems in a rigorous way. I will be presenting the step-wise refinement method (using Event-B) and its parachute methodology (introduced by J.-R. Abrial). Using this approach, users write an abstract model of a program and define properties over it, and then transform the model into implementation via a series of refinement steps (i.e. adding more details and properties to the system). Along the way, users need to formally prove the correct behaviour of the system. This technique is known as Correctness-by-Construction.
02.10.2015 - Larisa Safina

Implementation of a rich type system for microservices-based systems

Abstract:

Microservices became popular architectural style in the last few years. Jolie is a first programming language to work with them as with native features, so the main building block is not an object or a function, but a service. Current releases of Jolie lack of some useful primitives for message types and data manipulation, which would enrich its syntax and add further flexibility. In particular, Jolie does not support a choice operator for its message types. Type choices have proven to be useful in many other languages and are normally adopted by Web Services standards that Jolie has to deal with, e.g., WSDL.

The goal of this presentation is to describe the basic architecture and functioning of the Jolie interpreter and how it was extended by implementing a choice operator for message types



15.10.2015 - Leonard Johard

Strong AI in practice

Abstract:
New discoveries have narrowed the gap between biological and artificial intelligence. Machine learning has given us important insights into neural development, neural coding and the true meaning of dreams. Meanwhile, the study of dopaminergic neurons have inspired new approaches to robotics. Dr. Johard gives an overview of recent developments and shares his views on the future of neuroscience, robotics and the new IT industry.

Past sessions 2014-2015

Speech technologies. State and perspectives

14/05/2015 - Khusainov Aidar


30/04/2015 - Qiang Qu 

23/04/2015 - Joseph Brown 

09/04/2015 - Alexander Toschev

13/03/2015 - Giancarlo Succi

26/02/2015 - Dina Nasyrova (GDC) 
 

Functional and non-functional aspects of distributed systems
4 Sept. prof. Salvatore Distefano

Technological trends push towards higher standards often achieved by automation and computing systems, in particular distributed ones.This implies to properly consider both functional and non-functional aspects in their life cycle.Adequate techniques, methodologies and tools to manage as well as to represent, analyze and optimize dependability and performance of computing systems, especially when these latter are involved in (safety) critical systems are required.

This talk will discuss some solutions, techniques and tools for the management and evaluation of distributed systems.It is structured in two main parts: the first part focuses on functional aspects, mainly related to the management of new distributed system infrastructure derived by grid, volunteer, Cloud computing, sensor networks and cyber-physical systems; the second part deals with non functional properties such as dependability and performance, with particular reference to software performance engineering techniques, methods for the quantitative assessment of distributed system reliability and availability through stochastic models, considering non-Markovian phenomena and changing operating conditions.

For more information, please, see video Salvatore Distefano talk. 

Modern approaches to robust inference under heterogeneity, dependence, crises,structural breaks, outliers, and heavy-tailedness
25 Sept. prof. Rustam Ibragimov

We develop a new general approach to robust inference about scalar parameters of interest when the data is potentially heterogeneous and correlated in a largely unknown way, as is typically the case in different fields, including finance, economics, markets for informational technology products and technological innovations, and many other areas related to IT. 

The approach is based on small sample conservativeness properties of the standard one-sample t-statistic and two-sample t-statistic for testing equality of means. These properties show that, for commonly used significance levels, the one-sample and two-sample t-tests remain conservative for underlying observations that are independent and Gaussian with heterogenous variances. One might thus conduct robust large sample inference in the following simple way: partition the data into some number of groups, estimate the model for each group, and conduct standard one-sample or two-sample t-test with the resulting  parameter estimators of interest.

Hidden Markov Models Discriminative Training Methods for Automated Speech Recognition
9 Oct. prof. Edward Minniahmetov(KFU)

In general, speech recognition systems, based on hidden Markov models, train models for each sub-word unit (or for each word in a small vocabulary recognition system). In the recognition stage, the probability of the unknown utterance, given each one of the models, is calculated. The model with maximal probability defines the recognized word. 

The hidden Markov model is most often trained by the Baum-Welch algorithm, which is based on maximum likelihood criterion. Baum-Welch receives samples of each word (or sub-word unit) and, given a suitable model, tries to maximize the probability of each sample. Under the assumptions that the hidden Markov model is the true model of the utterance and that an infinite number of samples of each unit is available, this training method should give the minimal recognition error rate. 

However, in reality the two assumptions are not accomplished, so that Baum-Welch training fails to lead to the minimal recognition error rate. In the last 20 years new hidden Markov model training methods with the direct goal of minimizing recognition error rate have been developed. These methods are called discriminative training methods. There are three most popular discriminative training methods: Segmental General Probabilistic Descent, Maximal Mutual Information Estimation and Corrective Training.

Aliascalculus, change calculus and more
17 Oct. prof. Bertrand Meyer

Many of the most difficult and important problems of program analysis boil down to alias analysis: finding out whether two reference (pointer) expressions can become attached to the same object. Since 2010, with the help of colleagues (particularly Alexander Kogtenkov and Sergey Velder) I have developed an “alias calculus” and applied it to such goals as change analysis (also known as frame analysis) and deadlock detection. The talk describes this work and presents what can be achieved today in automatic alias and change analysis as part of the Eiffel Verification Environment.

For more information,please, see video Bertrand Meyer talk.

On a function counting smooth numbers
23 Oct. prof. Shamil Ishmukhametov (KFU)

Smooth numbers play an important role in Number Theory and Cryptography as an opposite to prime numbers. A natural n is y-smooth for a positive y if all divisors of n are less than or equal to y.  Let psi(x,y) be a function counting the quantity of y-smooth numbers less or equal to x. In our report we analyze algorithms for counting psi(x,y)  including a new one elaborated in Kazan and  give their applications.

Agile development
31 Oct. prof. Bertrand Meyer (IU)

Agile methods mesmerize the industry; they are a remarkable mix of the best and the worst. In my book “Agile! The Good, the Hype and the Ugly”, just published by Springer, I try to sort out the gems from the gravel. The talk pursues the same goal: enabling the listener to decode the agile buzz, stay away from awful agile advice, and benefit from the brilliant insight

For more information,please, see video Bertrand Meyer talk .

Affective Computation and Artificial Consciousness
13 Oct. MaxTalanov (IU)

How can we make machines actually feel emotions? Is there any option to make AI suffer, feel happiness, love, aggression, contempt, awe? Before we could find proper answer to this question we should take in account several aspects: philosophical, psychological, neurophysiological and cognitive architecture combine these theories.

For more information, please, see video Max Talanov talk.


Models for development of software applications
21 Nov. Arslan Enikeev (KFU)
This report presents an approach to the creation of a technique of the software applications development based on the integrated development environment which contains a model and tools for its implementation. The formal tools of a model specification and analysis are also considered. 

For more information,please, see video Arslan Enikeev talk. 

Reputation Management in Online Social Networks and its Applications
11 Dec. Jooyoung Lee (IU)
This talk focuses on a quantification of reputation and presents models which compute reputation within networked environments. Reputation manifests past behaviors of users and helps others to predict behaviors of users and therefore reduce risks in future interactions. There are two approaches in computing reputation on networks- namely, the macro-level approach and the micro-level approach. A macro-level assumes that there exists a computing entity outside of a given network who can observe the entire network including degree distributions and relationships among nodes. In a micro-level approach, the entity is one of the nodes in a network and therefore can only observe the information local to itself, such as its own neighbors behaviors.



Сайт находится в технической разработке