lessons learned from 2022
a thank you
To us, being the best machine learning summer school in health & biosciences is not the same as being the most perfect one. Rather, being the best means that the space we have created is not only open to all feedback, positive or negative, but also harnesses the feedback in a way that benefits everybody going forward. It is important that we learn from and take ownership of our mistakes.
In asking our participants for feedback - through surveys and conversations both during and after the event - we gain invaluable insight into what works and what doesn’t.
Yes, we leave ourselves vulnerable to criticism; to be receptive and open to it is scary, but in doing so, it helps us as a learning space to grow and evolve. Constructive criticism is necessary for identifying flaws in our ideas, and providing suggestions for improvement. So in reading, listening to and understanding our negative feedback, we gain a sense of clarity toward a more positive future, helping us to become an improved, more successful space for our participants in the years to come.
We would like to express our gratitude to all who provided feedback for MLSS 2022: whether that was through surveys or conversations, during or after the summer school, named or anonymous. In particular, we would like to thank those who provided us with open, honest and critical feedback; without those comments and suggestions, we could not have made the positive changes necessary as we progress into MLSS 2023.
moving forward
So what will be different about 2023? With all the feedback read, we have taken note of the aspects of 2022 that went well - and therefore we will be continuing or adding more of - and those that did not go so well - and therefore we will be removing or changing.
These points are summarised below:
we are continuing ...
General
We were pleased to hear that participants enjoyed the mixture of technical and design and communication skills sessions offered in 2022.
We will be continuing with this, and the sessions will follow the same structure as last year (lectures, tutorials, leadership conversations and communication workshops).
Lecturers
Our participants enjoyed the number and variety of lecturers in attendance of the ‘22 school. Participants enjoyed connecting with them in and out of teaching sessions, and hearing insights from both from their professional and personal lives. Also, having their details available in advance of the school on our website was well received.
Luckily for us, many of the 2022 lecturers will be returning for a second year at Bumblekite! On top of this, guided by feedback on which topics were missing last year, we will be also inviting lecturers with expertise in even more exciting fields.
"it was impressive how many different leaders were joining the sessions" - anonymous attendee, post-MLSS feedback
Engineering lectures
Participants felt that the lectures presented an interesting and wide variety of topics, and many felt inspired to go on and learn more about topics that they hadn’t come across before (e.g. genomics). Going forward, we will be continuing with the topics that were a success in 2022, and also updating our content with new topics suggested by our alumni.
Leadership conversation series
These sessions were generally the most well received, offering a unique and novel addition to the standard summer school curriculum. Participants enjoyed learning more beyond the technical expertise of lecturers, including perspectives on wider topics in academia and industry. People left feeling inspired and as if their own opinions had been challenged. Due to the success of these sessions, they will again form a fundamental part of the 2023 curriculum.
“What set the summer school apart were the leadership conversations at the end of every day where accomplished scientists and change makers held an honest and interactive discussion on ethics, leadership, and responsibility in our roles as data scientists.” - Muhammad Ali, MLSS ‘22 attendee, LinkedIn
“It was my great pleasure to be panelist during Bumblekite’s Machine Learning Summer School (MLSS) this week in Zurich” - Gorana Dasic, MLSS ‘22 panelist, LinkedIn
Office hours
The office hours were widely enjoyed by our participants, as they were able to ask all the questions they had, receive personalised advice for their own projects, and forge connections that were useful beyond the summer school.
The office hours will again be offered in 2023, and we are aiming to extend them to allow for a greater capacity of meetings to be made.
Participants
In 2022, we welcomed a wide range of participants from diverse academic, professional and personal backgrounds, providing everyone with fantastic networking, social and communication opportunities.
We strongly believe in the value of interdisciplinary collaboration, and so for this year, we will continue to strive to welcome a diverse group of participants.
“So many lovely people to meet and learn from” - Romana Burgess, MLSS ‘22 attendee, Twitter
Food and catering
Participants enjoyed the food options available in 2022.
In 2023, we will be continuing the same partnerships with the restaurants and coffee shops in the previous year, and hopefully with some new additions.
Feedback
Participants felt that their feedback - as collected via surveys throughout the week - was listened to and acted upon.
In 2023, we will continue to collect and address feedback before, during and after the summer school.
we have changed ...
Session topics
While the curriculum topics in 2022 were a success, the dynamic intersection of health, care and technology is constantly changing, and so must our content.
This means we will be updating our content for 2023, including new topics as suggested by our alumni. We will also aim to cover more machine learning fundamentals, in line with feedback from 2022.
Office hours
In 2023, the office hours will be held slightly later than in 2022 (i.e., not directly after the lecturer’s session). This will allow the participants to chat with the lecturer in a more informal, unstructured way, in advance of the office hours.
Technical prerequisites
Some participants felt that more technical expertise was needed (e.g. level of Python knowledge) as a prerequisite to the school in order to fully benefit from technical sessions.
Going forward, we will be more explicit about the level of technical expertise expected, and direct participants to technical resources to update their knowledge if needed.
Lecturers
Many participants suggested that lectures and tutorials given by a single lecturer - delivered one session after the other - were not as successful as those delivered by separate lecturers. It was suggested that delivering sessions in this way only gave participants a one sided perspective on a topic.
We agree with this, and for 2023, we plan to have separate lecturers per session throughout a single day.
Technical preparation for tutorials
One of the largest challenges we faced in 2022 was the organisation of the tutorial sessions, as often more time was needed for technical set up than anticipated.
Tutorials in 2023 will have clear datasets, notebooks and repositories available in advance of the sessions where possible.
Tutorial structure
We also faced structural difficulties in the 2022 tutorials, as there were challenges with downloading and accessing datasets, and the sessions would have benefited from more rigidity in their structure.
In 2023, we will better structure these sessions; each session will incorporate detailed guidance, expectations and end goals (deliverables).
“I'd like more briefings or conclusions on the tutorials, what am I supposed to be looking for? What methods should or could I be using?” - anonymous attendee, mid-MLSS feedback
Social programme
In 2022, many participants felt that the summer school didn't do enough to facilitate socialising. Whilst we had made plans for social activities in 2022, we unfortunately had to cancel them on account of the COVID-19 pandemic.
We will be putting a social programme in place for 2023.
Accommodation
Previously, we did not provide guidance on accommodation during the summer school.
This year we are working on creating partnerships with accommodation providers, in the hope that our participants will find themselves staying in the same location as one another.
Coffee breaks
During the breaks, we will be directing participants towards a larger selection of coffee venues. We hope that this will prevent queues, and allow participants to spend more time networking and connecting with one another.
Communication in practice
Some participants felt that they didn’t understand the value of our communication sessions.
We will be addressing this by adding a new session - “Communication in practice” - which will aim to answer this question.
Venue communication
We will be providing confirmation of the venue earlier this year, so that participants are able to organise their logistics (e.g., accommodation) as soon as possible.
Programme availability
We will be making the summer school programme available on the website at an earlier time point this year.
the data informing our changes
To approach this process of self-evaluation, we used a wide range of rich data collected from participants, lecturers and the Bumblekite team in 2022.
This included: 23 mid-MLSS questionnaires (9 anonymous), 8 post-MLSS questionnaires (3 anonymous), 85 completed daily reflections during the summer school (29 anonymous), post-MLSS interviews with 9 participants (each interview lasted around 30 minutes), project review notes from 4 Bumblekite team members, and unstructured conversations with the 2022 lecturers.