The 2018 Conference

Social Justice in an Age of Datafication

The Data Justice Lab is a new space for research and collaboration at Cardiff University. We investigate the social justice implications of datafication. In this talk, I discuss a number of the Lab’s ongoing research projects and argue more attention needs to be paid to data harms so that we can learn from where things are going wrong. By attending to data harms and those trying to prevent them, we learn more about how our present systems of government oversight are limited, as well as the need for greater transparency, accountability and means for meaningful citizen intervention into changing data practices.

https://www.cardiff.ac.uk/people/view/245594-

Deep Learning for Self-Driving Cars

Machine learning and AI is booming due to recent advances in deep learning and parallel computing. In this talk, we will consider opportunities and challenges with deep learning applied to the self-driving car.

Erik Rosén

Erik is a technical specialist and product owner of Deep Learning at Zenuity. Before that, he was the director of Automated Driving and Preventive Safety at Autoliv’s global research division (2014-2017). He has a PhD in superstring theory from Chalmers (2006) and eight years of experience from statistical accident research at Autoliv. In this role, he was active both within the perception and decision-making aspects of active safety systems.

@erikrosen_

https://linkedin.com/in/erik-rosén-b08538107

Experimentation Culture

In this talk Oscar will share how Booking.com, the world’s leading accommodation website, enables product teams to have more than 1,000 tests running at any given time.

Oscar Carlsson

Oscar is a Data Scientist at Booking.com’s HQ in Amsterdam.

https://www.linkedin.com/in/oscarlsson1/

Building a Large Scale Learning Machine

Artificial Intelligence and Machine Learning have made enormous progress during the last decade, and AI techniques are already widely used in industry and society. Expectations on these technologies are high, often so high that the technology we have today cannot come close to fulfil them. We will discuss what modern AI is, what is possible and what problems remain to be solved. We will present what RISE AI does within the area, how to create more interpretable machine learning models, approaches to bridge the gap between learning and reasoning, and what we believe is a way forward towards more general, large-scale AI-systems.

https://linkedin.com/in/daniel-gillblad-05b4632

Learning to Communicate using Deep Reinforcement Learning

The AI agents Alice and Bob walk into a bar. Alice wants Bob to order that new fancy drink everybody is talking about. Sadly, she has no word for the drink since it didn’t exist when her LSTM network was trained. She gets a pint.

Though, that story worked out just fine it points to one of the many problems with the current paradigm of learning language from canned text. In an alternative paradigm, that is currently gaining traction in the research community, agents instead learn to communicate by solving tasks, e.g. ordering the correct drink for their buddy and getting reinforced if it succeeds.

This leaves the agents with a functional language grounded in real objects and the ability to form new words when necessary. Further, the properties of the emerged artificial language can be used to answer importation questions regarding our own language development.

Mikael Kågebäck

Mikael is a last year PhD student in the Machine Learning group @ Chalmers, CSE.

https://www.linkedin.com/in/kageback/

Scoring Data

This talk will discuss strategies for producing a number that's valuable to your end users. Many models involve assigning a number or label to data, but how do you build and assess such a model if there's no accepted gold standard to measure against?

https://www.linkedin.com/in/simmonskristy/

Are Your Models Resistant to Adversarial Attacks?

The past decade has been marked by incredible advances in machine learning, especially in the area of computer vision. Increased interest in applying machine learning to safety-critical systems (such as self-driving cars) has raised the question of how secure these models are, especially when it comes to adversarial attacks.

An adversarial attack is when an attacker slightly modifies the input data in a way that deliberately causes the machine learning classifier to misclassify it while appearing unmodified to human observers. This poses a huge risk since attackers could for example target autonomous vehicles by using stickers to negate stop-signs in traffic.

This talk will focus on giving an introduction to adversarial attacks, providing historical and theoretical background. Different methods of both generating and defending your model from adversarial attacks will be discussed, as well as the current state of research in this field.

Marko Cotra

Marko is team leader & deep learning consultant at Berge. He has worked in the field of machine learning for two years. The main areas of interest include target tracking and vision-based object detection.

https://www.linkedin.com/in/marko-cotra-977b513a/

An Inspirational Map of West Sweden Big Data, Machine Learning & AI

(Co-presenter: Erik Behm)

Business Region Gothenburg has made a map of interesting actors in the Gothenburg Region in the areas of Big Data, Machine Learning, and AI. Close to 100 companies and organizations are listed as an inspiration. Get to know more about what’s happening in the field in West Sweden. Do you know the top three areas for AI in West Sweden?

@JohanHogsved https://www.linkedin.com/in/johanhogsved/

Making Data Accessible

Data may be abundant within your company, but to build powerful applications you need to make it flow in the right directions at the right time. This talk will be about lessons learned from building a data science team at Collector Bank, and our way forward.

Jonas Alm

Jonas Alm has been at Collector Bank since 2015 and is one of the creators of the Data Science team. He holds a PhD in Mathematical Statistics from Chalmers University of Technology and actively promotes more collaboration between academia and industry.

https://www.linkedin.com/in/jonas-alm-6780092/

Doing is Everything

(Co-presenter: Sebastian Nabrink)

From 0 to 100 in ten months. How Stena went from nothing to deployed machine learning models. Learn how we created a data science team with the ability to do everything from building and deploying models to communicating our results. Oh, and did we mention that the team only consists of junior colleagues?

https://www.linkedin.com/in/jessica-andersson-130945120/

An Inspirational Map of West Sweden Big Data, Machine Learning & AI

(Co-presenter: Johan Hogsved)

Business Region Gothenburg has made a map of interesting actors in the Gothenburg Region in the areas of Big Data, Machine Learning, and AI. Close to 100 companies and organizations are listed as an inspiration. Get to know more about what’s happening in the field in West Sweden. Do you know the top three areas for AI in West Sweden?

@ErikBehm https://www.linkedin.com/in/erikbehm/

Machine Learning in the Cloud

The third wave of cloud computing is here and advancements in the field of machine learning are being made every day. In this era of ever-evolving technology, there are a plethora of resources available to machine learning practitioners for handling their data and models. In this talk, we will focus on training and deployment of machine learning models in the cloud, in particular, Google Cloud Platform. We will touch upon the use of Google Cloud Machine Learning Engine and the workflow associated with the product. Throughout the talk, we will go from writing TensorFlow code to uploading the code to the cloud for training and to deploying the trained model.

Kevin Kyeong

Kevin is a technology evangelist and enthusiast who envisions commuting on a multirotor hoverboard guided by his virtual assistant. Having studied at the University of Waterloo in Canada, he built and raced solar cars in the outback of Australia. After reaching the 10,000-hour rule in development of electric vehicles, he has pivoted his career to software development in the field of artificial intelligence. His current interest lies in widespread applications of deep learning in the cloud. Today, Kevin works at Berge as both a machine learning practitioner and a cloud strategy consultant in an array of industries.

https://www.linkedin.com/in/kkyeong/

Determining the Health of Rotating Machines from their Vibrations

Combient is a joint venture with a mission to accelerate digital transformation within some of the largest traditional enterprises in Sweden and Finland. A major part of this work is carried out by our analytics team, consisting of a dozen data scientists, and has included predictive maintenance churn prediction, natural language processing and more. In this session, we will share challenges, methods and preliminary results from an ongoing project that Combient is carrying out together with SKF. We will describe a machine learning approach for determining the health state of rotating systems. The results are based on vibration data collected from 30,000 machines since 2001.

Doing is Everything

(Co-presenter: Jessica Andersson)

From 0 to 100 in ten months. How Stena went from nothing to deployed machine learning models. Learn how we created a data science team with the ability to do everything from building and deploying models to communicating our results. Oh, and did we mention that the team only consists of junior colleagues?

https://www.linkedin.com/in/sebastian-nabrink-9981a362/

How ML Revolutionized the way Self-Driving Cars Perceive the World

Self-driving cars started as a distant dream about 80 years ago. These days, this vision is becoming reality and self-driving cars promise to revolutionize our society by improving our current mobility models, and most importantly, by radically reducing the number of road fatalities.

Machine learning plays a central role in the perception and understanding capabilities that these vehicles require to correctly operate, not only under standard conditions, but also at the most unexpected situations.

This talk will be around the way self-driving cars perceive the world, the key role that Machine Learning plays in there, and the open perception challenges that remain to be solved to have truly self-driving cars revolutionising our society.

Sebastian Ramos

Sebastian is a team lead of deep learning at Zenuity, joint venture of Volvo Cars and Autoliv. Before that, he was pursuing an industrial Ph.D. in computer science, with main focus on deep learning for perception of self-driving cars, at Daimler AG R&D and TU Dresden. During his studies, he also spent time as a visiting student at TU Munich, research intern at Siemens AG R&D and research assistant at the Computer Vision Center of Barcelona. Sebastian obtained his M.Sc. degree in computer vision from the Autonomous University of Barcelona and his B.Sc. degree in electronic engineering from the National University of Colombia. His research interests are at the intersection of computer vision, machine learning and robotics with a special focus on self-driving technology.

https://www.linkedin.com/in/sebastianramosp/

Text Classification in the Legal Domain

Data analysis of legal texts, especially such that requires high precision, is not achieved over a few months of using random algorithms and relying on freely floating datasets with semi-trustable annotation quality. It requires a team of highly motivated people (preferably with diverse skills), high-quality datasets and a lot of time for experiments, failures, refactoring, testing, and development as well as heated discussions.

In this talk he will present how Seal have worked with Machine Learning, Natural Language Processing, and data analysis of legal texts and what are the difficulties they have encounter as well as the approach they have take to address them.

Svetoslav Marinov

Svetoslav Marinov holds a PhD in Linguistics from the University of Gothenburg. He has worked in the field of Natural Language Processing and Machine Learning for more than 10 years. He has supervised a number of Masters' students at the University of Gothenburg, Lund University and the Chalmers University of Technology, as well as presented at conferences in Linguistics, NLP, and Machine Learning.

https://www.linkedin.com/in/svetoslav-marinov-b25626/

Predicting Response Time in a Large-Scale Information Retrieval System

Meltwater’s media intelligence platform executes millions of search queries every day. An Elasticsearch based platform contains 40+ billion documents and provides search results and analytics.

Queries can be interactive (a human waiting on the result) and non-interactive (batch-like queries for asynchronous reports). While most queries take in the order of milliseconds, some queries can be expensive and take seconds or minutes to execute. These ‘slow’ queries have a detrimental effect on the user experience, by hogging machine resources and increasing wait times. A 10-millisecond query that waits 30 seconds creates a bad quality of service.

Accurately predicting the execution time of queries makes it possible to segregate slow queries into a separate machine pool, thus mitigating the negative impact of these queries. Historical query logs contain an abundance of data and features to train a machine learning model on. However, the problem domain is complex and the data is noisy. With thousands of queries running concurrently across more than 400 machines causing interactions that affect query runtime.

This talk will describe the problem and the challenges it poses from a machine learning perspective. It will describe the attempts to solve it, with a particular focus on techniques and algorithms that are useful when working with large-scale noisy data.

David Burke

David Burke has worked professionally with machine learning for the past 9 years. He has been part of teams that successfully applied machine learning techniques both in the online and TV advertising domains at Admeta/WideOrbit and is now applying his knowledge to problems in the information retrieval domain at Meltwater.

David has previously presented talks in several international workshops and conferences and more recently (Feb’ 2017) hosted a machine learning meetup at WideOrbit’s Gothenburg office.

Machine learning in information retrieval is an exciting area and one that Meltwater want to be at the forefront of. This talk will present knowledge on what we are doing on one specific project, but with potentially several more ML projects in the pipeline, we see Meltwater becoming an active participant in the Gothenburg machine learning community.

We want to share our problems and solutions in order to generate discussions, build relationships and continuously learn and improve in what we do.

https://www.linkedin.com/in/ekrubdivad/