…wait! Let’s re-scale it to leadership, teaching and business…
Web pages are currently under work.
I will publish a series of communication and leadership posts later in the spring of 2023. The machine vision series is continuing also, with illumination, HS imaging and machine learning posts as soon as I spare a few moments to write.
Anna-Maria defended her dissertation in December 2022. At the same time, she finalised her five-year-long journey of deepening her skills into a new career. Her computer science master studies started in January 2017 when she wrote her “Hello world” in C#, with no previous experience in programming.
Before her career change, she has been working in the business field, including versatile tasks in marketing, productizing, sales, accounting, customer service, business management, and HR. She has collected approximately 8 to 10 years experience of in leadership (maternity leaves affect the number of years). During her years as a team leader, she participated in several management and leadership programs provided by the companies she worked for. The work experience arises from SMEs, listed companies and the public domain (the Social Insurance Institution (SII)).
Currently, she is a PhD in computer science, with a wide experience and skills from sensor-level programming to data processing and analysis. During her studies, she was lucky to be hired to the Spectral imaging laboratory, where she has been working for nearly five years.
She started as a software developer, and she is the main author of the CubeView software, which is software for hyperspectral imaging and analysis for VTTs prototype imagers. She is capable of designing and implementing machine vision systems, from the device workflows enabling scientific spectral data gathering with customised user interfaces.
The gathered work experience during her master’s studies related heavily to designing and implementing device-controlling phases for machine vision sensors and optical components and user interface developing tasks.
After she graduated (MSc), she focused on computational data analysis, achieving Aaltonen Säätiö grant. She has been developing three new machine learning methods (New versions of Minimal Learning Machine, MLM, see Nokia awards 2022) and skin cancer research from hyperspectral (HS) images, using data that is captured with the CubeView Hospital imaging system, which she developed.
The skin cancer research was conducted using convolutional neural networks and 3D HS imaging. During her doctoral studies, she worked as a part-time machine vision engineer at Solteq Robotics (see the award-winning retail robot).
Currently, she continues her research in the field of computational data science (wildfire detection and prediction, FireMan project). She is a teacher of machine learning, anomaly detection and machine vision courses at the University of Jyväskylä. She is building a machine vision laboratory, a concept that allows students to get hands-on with machine vision sensors and Raspberry Pi cameras. She works as a supervisor for two master’s students and two research assistants and begins supervising a graduate student this spring. She enjoys teaching and actively tests and develops methods that are close to leadership methods, allowing the students to be active, and see and understand the meaning of the subjects related to their career dreams.
Aiming for academic or business-research collaboration? Feel free to contact! Click here to see Anna’s current contact information at the University of Jyväskylä.
Since I have education and skills in marketing, productising, sales, management and leadership, I wish for the future the chance to use these skills in my teaching, but to be the teacher that creates materials and teaches courses is not of the essence.
After the career change to Computer science and PhD studies, I feel that just becoming a teacher could be something I could have done without the research experience and five years of work. On the other hand, I have a good background, and besides the promising start as a scientist, I wish to combine these both.
My research career is at a good pace. Since 2019: Nine peer-reviewed articles, one completed and defended dissertation (includes six articles out of nine). I have worked for two months as a postdoctoral researcher, and so far, rour articles are in progress to be published before September.
I am supervising three master’s theses and the work of two research assistants, and in April 2023 I will start supervising a PhD student which is to be hired into my team in FireMan project. I am slowly building a machine vision laboratory infrastructure, which allows us to offer students and other research groups the possibility to use special machine vision systems and other sensors with Arduino and Raspberry Pi platforms in their studies and research. Therefore, the next career change should enable both, research and teaching. I wish for a tenure position, which allows me to use my management, leadership and marketing skills in developing curriculum and teaching besides working as a researcher. Then, based on my research, I could teach courses on a topic such as machine vision, tiny ML etc., where I am already bringing hands-on education with the machine vision laboratory.
I wish for a tenure position, which allows me to use my management, leadership and marketing skills in developing curriculum and teaching besides working as a researcher. Then, based on my research, I could teach courses on a topic such as machine vision, tiny ML etc., where I am already bringing hands-on education with the machine vision laboratory.
As a supervisor (and former adult student), I have noticed that a certain group of students that study and work at the same time benefit from certain tools. Time is one of the most valuable things when trying to complete a thesis, work and handle your daily life simultaneously. Therefore, it might be valuable to use tools for raising the effectiveness and for deleting time-related barriers between the started and finished thesis.
Since my master’s thesis was about joining scrum and design thinking, I started to think about whether we could use sprints, tasks and dailies and offer students possibilities to work efficiently close to supervisors.
A scrum framework is familiar to many students working in software engineering. I took the liberty to join elements of scrum, coaching, and sparring, and as a result, we are now piloting Thesis Sprints at the Spectral Imaging Laboratory. The method offers tools for fluent thesis work.
In our sprints, we invite students to work in the office for a short period. Before joining the office, the student will plan the sprint’s goals and tasks and divide them into tickets. Each morning, we will meet and discuss the planned daily tasks. The student can ask for help at any time (since we spend time in the same office, and have lunch and coffee breaks at the same time; it is easy to ask small things that might take time in traditional email-based supervision). After each day, we spend a few minutes to retrospect and pay attention to what is accomplished and select the tickets for the next day. After the pre-agreed office time, we have the final retrospective, where we can discuss face-to-face the next steps. The last retrospective aims to help students to continue using tasks and tickets while performing tests and writing the thesis.
Planning
Sprint!
Lessons learned from Thesis Sprints
Here are examples of strategic teaching development in which I have been participating
My positions in the University collegium and the faculty council offer me a larger strategic view and understanding of teaching and research policies and their development in our University.
1.1.2022-31.12.2025 Member of the faculty council (elected)
1.1.2022-31.12.2025 Member of University collegium (deputy, elected)
Besides the strategic council and collegium work, I have been actively participating in strategic teaching development in our faculty. The main topics for the years 2022-2023:
2022
2023
Besides conferences, I have been presenting our research as a visiting lecturer and conference presenter.
Feel free to contact me, if you wish to have a lecture, speech or presentation. Example topics: hyperspectral imaging, machine vision, computer vision, machine learning etc.
Click here to see my current contact information at the University of Jyväskylä.
2022 ISPRS International Society For Photogrammetry ad Remote sensing: Updating strategies for distance-based classification model with recursive least squares. Awarded with a live presentation, pre-recorded presentation and poster presentation.
DIME21 Eccomas Thematic Conference, JAMK University of Applied Sciences, Jyväskylä, Finland. FPI-based hyperspectral imager for the complex surfaces -toward optical biopsy
2021, ISPRS International Society For Photogrammetry ad Remote sensing: Piecewise anomaly detection using minimal learning machine for hyperspectral images. Remote participant, pre-recorded presentation
2020, SPIE Image and Signal Processing for Remote Sensing XXVI: Minimal learning machine in hyperspectral imaging classification. Remote participant, pre-recorded presentation
TIES411: Konenäkö ja kuva-analyysi
Target: Master-level students.
My status: I am one of three responsible teachers, developing course materials, lecturing and supervising tasks, and responsible for the hands-on demos.
Course level: Advanced
Course actions: This course has a dual approach. Mathematical foundations are introduced and applied into practice using Python. At the same time, students are learning how to work with machine vision sensors, and control devices, to capture and modify the captured images. The demos with Basler sensors and Raspberry Pi cameras deepen the learning experience. The demos will be live.
Exam or demos (5 ECTS)
Learning outcomes: Students understand how to design and build machine vision systems. A student can capture and modify the data and understands the mathematical-physical limitations of machine vision. Students will learn to extract features, work with 3D data, and apply multiple mathematical functions to real-life image-analysis tasks.
Teaching challenges: The course format is new, rolling in period 4 / 2023. The course material is currently under work. This course will be developed within the next years.
Goal – in future, this course could be flipped and materials enlightened with videos
KYBS3050: Koneoppimismenetelmiä kyberturvallisuuteen
Target: Master-level cyber security students.
My status: I am one of two responsible teachers, developing course materials, lecturing and supervising tasks.
Course level: Advanced
Course actions: At first: Python, Jypyter notebook & Sckit learn: Building the environment, testing some classification, clustering and anomaly detection methods for warming up.
We introduce basic machine learning methods and learn how to use them in cyber security issues: Curse of dimensions, SVM, Time series, NLP, MLP, CNN, Autoencoder, CNN autoencoder, Reinforcement learning, transfer learning, GAN, adversial attacks, Minimal Learning Machine.
Demo + essay (5 credits)
Learning outcomes: A student understands machine learning principles and can apply machine learning to various issues related to information security.
Teaching challenges: The students have varying backgrounds; for example, maths is not required. Therefore the course has been traditionally lecture-based, and we provide personal guidance for exercises.
Goal – in future, this course could be flipped and materials enlightened with videos
KYBS7041: Anomalian havaitseminen
Target: Master-level cyber security students.
My status: I am one of two responsible teachers, developing course materials, lecturing and supervising tasks.
Course level: Advanced
Course actions: First: Python, Jypyter notebook & Sckit learn: Building the environment.
Introduction to anomalies and categories. Review of required linear algebra and math skills. Eigenvalue decomposition and singular value decomposition. Data types and data pre-processing and processing. Method evaluation and validation, thresholds, clusters, and neighbour-based methods.
Exam (3 ECTS) or demo + exam (5 ECTS)
Learning outcomes: Students will understand the concept of anomaly and the different categories, concepts of data point, data set and type and different categories. Students will learn the operating principles of different anomaly detection methods and can apply anomaly detection methods in different situations. Can evaluate the performance of methods.
Teaching challenges: The students have varying backgrounds; for example, maths is not required. Therefore the course has been traditionally lecture-based, and we provide personal guidance for exercises.
Goal – in future, this course could be flipped and materials enlightened with videos
The teaching strategy arises from my own experiences as an adult Master’s student without previous computer science background (and with a full-time job, big house, two dogs, two kids, three lambs and so on…). When participating the basic teaching with young co-students, there were definetly many situations to learn from the pedagogical point of view. Similarly, my eight years of business experience with leadership education (MBA) completed with management courses and coaching experience gives me perspective to season my teaching.
I have been teaching now for two years as a shared principal lecturer in two courses (3-5 ECTS and 5 ECTS, listed below). Currently, I am preparing a third course, which enables me the role of one of three principal lecturers and the possibility to design the course materials and utilise flipping. I studied my pedagogical studies in 2022, and since then, I have been actively developing the following goals and tools for teaching.
My teaching goals
As a teacher, my personal teaching philosophy is simple. I believe in open communication, good materials and easy to approach atmosphere. In every course, instead of filling the students’ minds with mountains of information, the aim is to deliver 5 to 6 main outcomes, the ones they will remember when woken up at 3 AM. These will be covered with details, but the aim is to deliver deeper learning experiences for the main topics and broaden the view over important things to know but not essential to spend hours with -they can later on go find the sources and deepen the understanding if necessary.
My way of working is visual. As a student, I need frames to fill with small details. Therefore, I often use figures to deliver information and deepen the understanding through the theories, which can be written, and lectured in front of students in the classroom or via videos.
Above is an example of the visualisations I have drawn for the students (Figures 1 and 2). It reflects my teaching statement, which is built from the strategy:
If you wish to test your idea of target-wisely customised imager, it can be built combining a machine vision sensor with optical or optomechanical components. There exists various of components to choose from.
Extension tubes (Figure 1) are useful for creating more complex prototype imagers than a basic sensor and a lens. With extension tubes, the optomechanical components can be mounted together.
Different band-pass filters (Figure 2), beam splitters (Figure 3) and lenses can be used to produce, for instance, a simple two-channel spectral imager.
While designing such a system, it is valuable to understand that the system’s parameters will change depending on where the extension is added. For example, suppose the extension is placed between the lens and the sensor. In that case, the image-side focal length increases, decreasing the minimal working distance and field of view, magnifying the object (Greivenkamp 2004).
As an example of special imager, a spectral camera typically consists of an optomechanical components, lenses and a sensor. The optomechanical component can be a prism or mechanical, dispersing the incoming light to wavelengths. We will later discuss over the spectral imagers and spend some time getting known with the dispersive components and their operating principles.
Found something useful? Wish to cite? This post is based on my dissertation. For citations and more information, click here to see the scientific version of it.