Machine Learning

Three-year pro­ject on test­ing of ma­chine learn­ing sys­tems launched

New research aims to develop methods and tools for the creation of safe and reliable AI systems. An interest group for Finnish companies is also planned.

VTT Technical Research Center of Finland and the University of Helsinki are, together with six Finnish companies, starting a three-year research co-operation in the context of the ITEA IVVES  (Industrial-grade Verification and Validation of Evolving Systems). Business Finland supports the project with 1,3 M€ for the University of Helsinki and 1,4 M€ for VTT.

The aim of the project is to develop methods and tools for the creation of safe and reliable AI systems.

“We can now significantly extend our research efforts on testing, continuous development, and maintenance of machine learning systems together with our partners in Finland, The Netherlands, Sweden, and Canada. We are also planning to set up an interest group for Finnish companies interested in the project,” says professor Jukka K. Nurminen from the department of Computer Science at the University of Helsinki.

“Making sure that AI solutions are, and stay, unquestionably reliable in real-life applications, also long after they have been installed, is an important topic that requires efforts - IVVES gives us the means to develop tools to address this properly” adds research professor Mark van Gils from the Data-Driven Solutions research area at VTT.

The University of Helsinki's work in the project is jointly headed by professors Tommi Mikkonen and Jukka K. Nurminen. At VTT, the project is headed by Johan Plomp (the country co-ordinator for the Finnish consortium) and research professor Mark van Gils.

Empirical Software Engineering research group

Artificial Intelligence at VTT

Update on 11 March: We’ve updated this news item by adding information about VTT’s role in this research project and including Mark van Gils’s comment.

”Machine learning excited me before I knew it was a thing”

Wherever you end up in working life, some kind of social skills and a capability to work together with people are needed, says Arno Solin. Photo: Matti Ahlgren / Aalto University

Wherever you end up in working life, some kind of social skills and a capability to work together with people are needed, says Arno Solin. Photo: Matti Ahlgren / Aalto University

Arno Solin is excited about machine learning, as it offers a way to both delve deep into theory and solve real-life problems.

Dr. Arno Solin, Assistant Professor, stores a plastic bag full of electronic gadgets in his office wardrobe, since a student of his needed equipment for building a robot. Part of them have come a long way. "My father bought this soldering iron. I was too little to be in school at the time, or just barely in elementary school," Solin says and laughs.

Already as a child, Solin was interested in technology, physics and mathematics. At home, he would build robots and spaceships. The son of academicians, he saw firsthand the life of researchers – and wanted to become one himself. "After high school, I almost chose Political History as my major. But then I figured it would be easier to have history as a hobby than statistics and mathematics."

In addition to working as an Assistant Professor at Aalto University, Solin is a researcher of FCAI. FCAI is collaborating with the Alan Turing Institute, based in London, and Solin is taking part in this collaboration.

Solin thinks that FCAI’s power lies in the fact that together researchers are stronger. Individual researchers typically need to market their research on their own, but a background institution – such as FCAI – helps to make AI research more concrete for decision-makers and non-professionals. It also helps to build the brand of Finnish research internationally.

“Development of methods isn’t just about doing something at the university, but instead, there’s a greater background idea and clear goals in the long run.”

 

Machine learning combines theory with solutions to tangible problems

Solin studies machine learning, which is an application of artificial intelligence: the machine learns based on experience without being further programmed by humans. Machine learning makes use of statistics, for example.

In June 2019, the Academy of Finland granted the research project led by Solin funding allocated to the new generation of researchers. Researchers make use of statistical machine learning and the development of computer vision in their project entity.

“Development of methods isn’t just about doing something at the university, but instead, there’s a greater background idea and clear goals in the long run.”

Solin finds machine learning fascinating, as you get to combine theory with solutions to tangible problems. He concentrates on probability modeling: How do you model uncertainties? How does machine learning deduce results from new data? How can you help machines reach sensible deductions in the here and now?

"I think I was interested in machine learning already before I knew it was a thing."

In the Academy-funded project, researchers concentrate on sensing, comprehending and describing the environment via machine vision methods. These functions are a challenge in the development of any autonomous or augmented reality system, especially when the surrounding conditions are uncertain.

The project has the potential to develop methods that could solve many a practical problem. This potential fascinates him. New research results could help develop, for example, the functionalities of smartphones. Computation, especially, can be used to make them work better, to use the current data more efficiently. Smartphone cameras could become better for shooting at night, or provide better depth of field without bigger, better and more expensive sensors.

"Existing sensors, existing smartphones can improve and have more to offer simply by being able to arrive at conclusions more effectively from information detected by the device," Solin sums it up.

New knowledge can be applied to many other things. For example depth estimation can help create video games, or devices for the visually impaired to better grasp their surroundings. The research is mostly pure research in nature, but provides reliable and efficient methods for the needs of other disciplines. Through collaboration, they have been adapted in medicine and the evaluation of urban air quality.

 

The researcher must know how to communicate

Solin, Assistant Professor since 2018, is the co-author of a textbook in stochastic differential equations together with Simo Särkkä, Professor of Electrical Engineering, and he has taught several courses at Aalto University as well as Introduction to AI at the Open University.

Solin laughs when he claims his motives for sharing knowledge are partly self-interested. "Explaining to others, I learn myself. When you have to explain things from different angles, in different ways and afresh, you get a different take on familiar things."

There's no point in being an ace researcher, if you are not able to communicate your findings to others

Solin considers teaching and conveying your own particular expertise to others as an essential part of a researcher's work. It supports research and raises new generations to study and apply what they have learned.

Already as a child, at the Waldorf School of his native Turku, he learned to present things visually and with clarity, and to hold presentations. He got used to going to some trouble to make things clear to his audience.

Photo: Matti Ahlgren / Aalto University

Photo: Matti Ahlgren / Aalto University

"That is something I think should be valued more. There's no point in being an ace researcher, if you are not able to communicate your findings to others."

The Waldorf School has strongly shaped the kind of adult and researcher Solin has grown up to be.

In his opinion, the pedagogy supports the pupil's personal growth. Studies advance in the pupil's terms. Group sizes are small, which enables tuition that is more personal. The school encourages self-expression and social interaction.

"Wherever you end up in working life, some kind of social skills and a capability to work together with people are needed."

Solin hears people are amazed at how many international partners he has. He puts it down to his proclivity to work with other people. "Maybe it comes through, and then others like to work with me, too."

Authored by Anu Haapala / Aalto University
English translation by Susanna Bell

Machine learning in chemistry - algorithms help finding minimum energy paths and saddle points more effectively

Reproduced from J. Chem. Phys. 147, 152720 (2017), with the permission of AIP Publishing.

Reproduced from J. Chem. Phys. 147, 152720 (2017), with the permission of AIP Publishing.

Machine learning brings new possibilities to many areas of research - also chemical research, which is shown in a recent dissertation written at the Finnish Center for Artificial Intelligence.

Olli-Pekka Koistinen, doctoral candidate at Aalto University, developed machine learning algorithms based on Gaussian process regression to enhance searches of minimum energy paths and saddle points, and tested how well the algorithms work.

In theoretical chemistry, finding minimum energy paths and saddle points is one of the problems that consume most time and computational resources. The bottleneck of the computation is the accurate evaluation of energy and forces for each atomic configuration, which typically needs to be performed at hundreds of points in the configuration space.

Algorithms utilizing machine learning can reduce the number of observation points and expensive energy evaluations to a fraction of what is required by conventional methods and thus speed up the computation.

Minimum energy paths lie on a potential energy surface that describes the energy of a particular system - a molecule, for example - in terms of particular parameters. Usually these parameters show the locations of the atoms. The local minimum points of the energy surface correspond to the stable states of the system. The minimum energy paths connect these points and describe possible reaction mechanisms.

“As an orienteer, I see this energy surface as a map. The stable atom configurations are shown as depressions on the map, and the minimum energy path is a route between two such depressions. It stays as low as possible all along the way. The highest point of the path is at a saddle point where you can get from one depression into another one staying as low as possible,” Koistinen explains.

Traditionally, researchers have searched for minimum energy paths and saddle points using iterative methods that proceed on an energy surface with small steps. With the help of machine learning and statistical models, previous observations can be utilized to model the energy surface, and the goal can be reached with significantly fewer iterations.

Therefore, machine learning offers a more effective and lighter and thus also cheaper and more ecological option. It can also open new possibilities for studying problems that have not been feasible with traditional methods. “This is another example of a research topic in which machine learning methods can be helpful,” Koistinen says.

Helsinki and FCAI will host a new ELLIS unit for top AI research

Finnish artificial intelligence research received a significant acknowledgement. Finnish Center for Artificial Intelligence FCAI will host one of the new European units of top AI research, as the European Laboratory for Learning and Intelligent Systems will establish one of its first units in Finland.

The first ELLIS units were announced at the ELLIS assembly on 10 December as a part of the NeurIPS 2019 conference. From left to right: Yoshua Bengio, Bernhard Schölkopf, Nuria Olivier, Matthias Bethge, Max Welling, Director of FCAI Samuel Kaski, Se…

The first ELLIS units were announced at the ELLIS assembly on 10 December as a part of the NeurIPS 2019 conference. From left to right: Yoshua Bengio, Bernhard Schölkopf, Nuria Olivier, Matthias Bethge, Max Welling, Director of FCAI Samuel Kaski, Sepp Hochreiter, and Yann LeCun.

ELLIS is a pan-European effort initiated in 2018 to secure the excellence of European machine learning research. It aims to ensure that Europe continues to be competitive with big economies, such as the US and China, and benefit from the newest findings of AI research. 

With the units, ELLIS wants to strengthen European AI research and collaboration between European researchers.

The unit will be founded in Aalto University and the University of Helsinki and hosted by the Finnish Center for Artificial Intelligence FCAI. Samuel Kaski, the Director of FCAI and Academy Professor at Aalto University, sees this as an excellent opportunity to boost basic AI research, which is the basis of all AI-related applications and impact. “Finland is very strong in AI research, and this new status is one indication of that.”

Professor Kaski believes that the ELLIS unit helps Finland to maintain its position as an attractive destination for top-level international researchers. It also gives current AI researchers in Finland more reasons to stay.

ELLIS aims to offer European researchers outstanding opportunities to carry out their research in Europe, and to nurture the next generation of young researchers in the important field of AI. All ELLIS units will arrange visits and events as well as provide funding for doctoral students in the ELLIS PhD programme.

The other cities selected to host a unit are Alicante, Amsterdam, Copenhagen, Darmstadt, Delft, Freiburg, Linz, Lausanne, Leuven, Oxford, Prague, Saarbrücken, Tel Aviv, Tübingen, Vienna, and Zürich.

Read more
Press release at the ELLIS Society homepage

Further information
Samuel Kaski
Professor, Aalto University
Director, FCAI
Phone +358 50 3058 694
samuel.kaski at aalto.fi

Click photo to see more photos from the ELLIS assembly.

Designing AI that understands humans’ goals better

Photo: Matti Ahlgren / Aalto University

Photo: Matti Ahlgren / Aalto University

To make a better smart assistant,  we need an AI that understands its user and does not constantly need detailed instructions

When researchers design machine learning systems, their goal is typically to automate certain functions. Instead of being fully autonomous, however, most of these systems work together with humans. In order to be truly helpful, they need to understand what goals people have.

Researchers at the Finnish Center for Artificial Intelligence (FCAI) have now taken important steps towards designing AI that understands people.

At first, the researchers taught the AI to build a model of its user - human or machine. Then, they taught it to adapt this model by following the user’s actions. In practice, the researchers developed machine learning methods which combine statistics with computation, and then tested the methods in practice and in simulations. They tested the algorithms in simple situations in order to make sure they understand what exactly happens in those situations and report about the events accurately.

In the first experiment, they designed an AI teacher for the learning AI.

‘This was difficult especially because the learning AI could decide what it wanted to learn,’ explains Samuel Kaski, the director of FCAI and professor at Aalto University. The researchers noticed that the AI learner achieved better learning results when the teacher understood what information the learner had already learned and adapted its teaching material to suit this  particular learner.

In the second experiment, human users were asked to find a particular target word by using an AI-based word-search engine. The engine presents the user one word at a time, and the user then  tells it whether the presented word is useful in finding the target word. If the user is looking for the word ‘football,’ for instance, they are likely to say that the first presented sport-related word is useful, if all the previous words have been related to food.

The results of this experiment showed that the AI could help the users in finding the target words faster if it understood that, by responding to the presented words in a certain way, the user wants to direct the AI towards the right words. In other words, the AI took into account the fact that the user is trying to teach it.

According to Professor Kaski, this topic is important, as the interaction between user and AI becomes much easier when the AI understands its user’s goals. ‘Then the human user does not need to explain in detail anymore what they expect from the AI helper.’

One of the main goals of FCAI is to develop AI that understands humans and is understandable. ‘So far, we can build AI systems that understand the users’ goals only in very simple situations, which means that designing truly helpful AI assistants calls for a lot of additional work,’ Kaski says.

The research article will be published at NeurIPS, the world’s largest and most prestigious machine learning conference that takes place in Vancouver, Canada on December 8 through December 14, 2019.

Link to the research article: https://aaltopml.github.io/machine-teaching-of-active-sequential-learners/

 

Further information

Samuel Kaski
Professor, Aalto University
Director, FCAI
Phone +358 50 3058 694
samuel.kaski@aalto.fi

Tomi Peltola
Postdoctoral Researcher, Aalto University, FCAI
Research Scientist, Curious AI
tomi@cai.fi

Finnish Center for Artificial Intelligence (FCAI) is a nationwide competence center for Artificial Intelligence in Finland, initiated by Aalto University, University of Helsinki, and VTT Technical Research Center of Finland. FCAI’s mission is to create a new type of AI that can operate with humans in the complex world and renew the Finnish industry. FCAI is one of the research flagships of the Academy of Finland.

Read also: Ten research papers from FCAI accepted to the prestigious NeurIPS conference

Every­one has their secrets – ma­chine learn­ing needs to re­spect pri­vacy

Photo: Susan Heikkinen / University of Helsinki

Photo: Susan Heikkinen / University of Helsinki

How can we teach artificial intelligence to make unbiased decisions? How can we protect citizens’ privacy when processing extensive amounts of data? Questions such as these need answers before the application of artificial intelligence and machine learning can be extended further.

In spring 2018, inboxes filled to the point of frustration with messages from businesses and organisations announcing their measures related to the entry into force of the EU’s General Data Protection Regulation, or GDPR by its common name.

The purpose of the regulation was to improve the privacy of citizens whose personal details are stored in various databases, a matter that is closely related to the research conducted by Associate Professor Antti Honkela at the University of Helsinki. At FCAI, Honkela heads a research programme focused on privacy-preserving and secure artificial intelligence.

He is specialised in machine learning that preserves privacy.

“Machine learning and artificial intelligence work best on massive repositories of data. These data often contain personal information that needs to be protected. The utilisation of machine learning must not jeopardise anyone’s privacy,” Honkela states.

Ap­plic­a­tion po­ten­tial in di­verse fields

Among other fields, machine learning could be used in medical research where extensive registers that contain medical records are employed as research data. Honkela himself has contributed to developing privacy-preserving machine learning techniques in targeting treatments to serious diseases.

“The aim has been to find a form of treatment best suited to individual patients. The same cure does not necessarily always work on a different cancer even though it might appear similar. The genome, for instance, can have an impact on the efficacy of various drugs. We have been developing techniques with which to work out answers from large datasets,” Honkela explains.

The utilisation of machine learning must not jeopardise anyone’s privacy

To have sufficient data at their disposal, researchers must be able to convince people that their research does not put the participants’ privacy at risk. It must be impossible to link sensitive information with individuals.

This is a problem Honkela is solving by developing methods in machine learning and statistics.

There is demand for machine learning that considers privacy also outside medical research. Potential for applications can be found in almost all areas of life, such as applications needed for research in various fields of science, the development of predictive text for mobile phones or banking systems.

We all have secrets

These days, the protection of privacy is a common topic. Honkela believes this is exactly as it should be.

“This is about a fundamental right. The Universal Declaration of Human Rights itself specifies that each human being has an inviolable right to privacy,” he notes.

Honkela says that society’s overall ability to function is based on people having secrets that stay safe. “Someone who says they have nothing to hide hasn’t thought it through,” he adds.

For instance, they could consider whether a company that provides medical insurance should have access to the genome data of its clients. Or whether a business looking to recruit new employees should be able to read personal messages written by applicants.

Those living in Western democracies may find it hard to grasp the potential consequences of a totalitarian state gaining access to the private data of their citizens.

Ma­chines must not dis­crim­in­ate

In addition to solving problems related to privacy, the broader application of machine learning in various sectors of life requires that consideration is given to how to make artificial intelligence unbiased.

“If machine learning is used in decision-making, we have to be certain that it doesn’t discriminate against anyone subject to those decisions,” Honkela points out.

Examples of discriminative decisions made by artificial intelligence have already been seen. Amazon, the online shopping giant, started using artificial intelligence to support staff recruitment. Eventually, the system was found to discriminate against female applicants.

“Perhaps previous data were used to train the machine. If more men have been hired earlier, the system may have interpreted this as something being wrong with women,” Honkela speculates.

Equality would also be a key feature in credit decisions made by banks or in granting various social welfare subsidies.

“For us researchers, there is still a lot to do in terms of the non-discrimination principle. For the time being, we haven’t reached a consensus even on the theoretical level on how to integrate it with machine learning,” says Honkela.

Authored by Anu Vallinkoski / University of Helsinki

When physician and AI work together, the patient benefits

The machine learning method by Iiris Sundin and her colleagues taken into account uncertainties of the world. This is important in decision-making. Photo Matti Ahlgren / Aalto University

The machine learning method by Iiris Sundin and her colleagues taken into account uncertainties of the world. This is important in decision-making. Photo Matti Ahlgren / Aalto University

Doctoral student Iiris Sundin learned in her studies that a machine learning model could make use of a physician's silent knowledge which usually is never written down. This kind of model predicts best how a given patient will react to specific treatment.

Artificial Intelligence opens up new avenues into health care, for example, but its potential is as of now not fully put to use. There are many reasons, but the most important one became clear to Iiris Sundin as she was starting her doctoral studies on machine learning: machine and Man must learn to work together.

"When I acquainted myself with the research of my advisor, Professor Samuel Kaski, on user modeling where the machine tries to understand the human, I realized the huge potential. Cooperation can mean other things than the human operating Excel or passively staring at projections on the screen," says Sundin, a doctoral student at the Department of Computer Science at Aalto University.

Samuel Kaski is the director of FCAI. Sundin's research combines Medicine, technology and many other interests of hers. She has also always wanted to work at something that benefits other people. On the other hand, she was interested in mathematics and programming.

The role of AI and machine learning in health care is researched a lot. Sundin's point of view is unique in having the machine make use of the doctor's knowledge to define the best possible care for the patient.

 

There are uncertainties in the world that must factor in when decisions are taken

Physicians have vast quantities of knowledge that is never written down and that is impossible to feed directly into a learning algorithm. Sundin and her colleagues have found out just how the machine could make use of such knowledge in e.g. figuring out the efficacy of a given type of medication.

Sundin and her colleagues have at their disposal, for example, data on gene specimens collected from cancerous cells, courtesy of FIMM, Institute for Molecular Medicine Finland. Researchers took a look at the mutations in the specimens and tried to predict which cancer medicine would work best for each patient.

We don't just discuss the average person, we acknowledge there are differences, and take it into account when reaching decisions.

It is crucial to remember that there are always uncertainties in the world, and thus, also a physician's knowledge of the effect of different mutations can be uncertain. The machine learning model devised by the research group takes this into account. The results show including expert knowledge in machine learning models and emphasizing data improve prognoses on how a given patient will react to a particular treatment.

So the model developed by researchers depicts realistically how sure people are of the different properties of real matters. "I would that such thinking were more common: we don't just discuss the average person, we acknowledge there are differences, and take it into account when reaching decisions. That way, things can be modeled in a more useful manner."

 

The researcher gets to tell something new about the world

Even though Sundin was fond of mathematics already as a child, and it comes naturally to her to see the world via mathematical thinking, art held a big role in her life, especially when she was young.

As a schoolgirl, she played the piano, sang in a choir, read books and attended art school. Now, she is into yoga, air acrobatics and camping. A researcher can be interested in many things, and creativity, for example, helps at work, too. "When you e.g. make posters for a conference, it is very useful if you can do some of the visual elements yourself."

The antics of engineering students were something Sundin grew up with. Both her parents graduated from the Helsinki University of Technology, and the family always celebrated First of May with parents’ college pals. All the adults sported the traditional tassel caps of engineering students. So engineering studies came naturally. "The degree of an engineer is really well-rounded and offers a good start. You can choose to be a researcher, industrial work or pretty much anything."

I realized researching was such great fun that I rather do more of the same.

Sundin completed her Master’s in Automation and System Technology. Although she never fancied herself a researcher, her interest in it was kindled when she was working on her dissertation. Sundin was modeling the properties of a drop of fluid on different surfaces.

Research offered a way to bring received mathematical ways of thinking and skills into the real world, and to tell something about that world. "I realized it was such great fun that I rather do more of the same."

 

”You succeed at work if you are open”

Sundin says she sometimes imagined researchers worked alone in their chambers. That is no longer the case, anyway. "You succeed better at work is you are open, socially adept and like to travel to conferences and network with the people there."

Research work is social. "You succeed better at work if you are open, socially adept and like to travel to conferences and network with the people there,” says Iiris Sundin. Photo Matti Ahlgren / Aalto University

Research work is social. "You succeed better at work if you are open, socially adept and like to travel to conferences and network with the people there,” says Iiris Sundin. Photo Matti Ahlgren / Aalto University

It is crucial for a doctoral student to possess the wish and motivation to understand things. The point of further studies is to delve into one fairly narrow area of learning very profoundly, something which can be demanding. Therefore, it is important to have tenacity and the ability to stick to it. "It is needed, if you want to make it to the end."

Even though Sundin has hitherto worked in basic research, she feels the methods developed need to be taken to a practical level, once the foundation laid by basic research is robust enough.

In the future, she hopes to apply what she has learned and continue work for the benefit of people and also the environment. "I hope to work at some Finnish institute doing research that help Finland make smarter decisions when it comes to the environment."

Authored by Anu Haapala
English translation by Susanna Bell