New in ML: A Guide for Navigating the ML Conference Scene
🔖 Tldr: Together, Khimya Khetarpal and Cheng Soon Ong go deeper in this blog on building a resource for new and early career researchers when attending a Machine Learning (ML) Conference.
Intended Audience
This blog post is targeted at people who are (almost) first time participants, whom we are calling "New in ML". It is most suited for folks who might fall in one of the following categories:
- First AI or ML conference
- First paper at the conference
- Looking for opportunities
- Finding mentors
- Connecting with peers: potential future collaborators
New In ML: Attending your first in-person Machine Learning conference?
Goals of conference participation
Considering you are New in ML, to make the most of this time when attending in person, we list a few of our personal recommendations as potential goals of your attendence.
[I] A core component: research
A core and large part of attending a ML conference is about research. Centered around research, the attendees have goals ranging from presenting their own research, to attending talks, to meeting researchers who might have overlapping research interests, to finding researchers with similar backgrounds or aspiring trajectory paths, etc.
Skimming through posters, accepted papers beforehand is useful to know what topics the conference broadly casts a net over. If you find related and relevant work to your current projects, this might look scary at first, and give a sense of insecurity. However, this implies more than one lab is investigating this particular problem, it is indeed an open problem worthy of studying, and that you have competition as well, which is healthy and adds to your research.
[II] Advertise your research, advertise yourself
If you are presenting a paper (either oral or poster), while it is true that the work will speak for itself, we highly encourage you to advertise your work! You have to be your own hype person! If your advisor or any collaborator is attending you could also encourage them to advertise the work. You are also advertising yourself (both your skills and also your ability to work in a team) as an interesting colleague or collaborator. You may be representing the place you come from, and the place you are currently working at.
But how? There are several ways to do this.
- Social media offers a low entry, no barrier channel to share your research. So find your favorite one and do advertise your work there.
- Prepare an elevator pitch about your work. Be ready to give this pitch anytime you meet and interact with someone!
- Prepare an elevator pitch about yourself as well. Remember this is an advertisement about you too, and its ideal to have a short icebreaker introduction for yourself. A simple example: Hi, I am Amanda, I am a PhD student at University of Netherlands. I work on reinforcement learning theory. I would love to chat more about your current research interests.
- Don't hesitate to say hello to people you see and recognize. Introduce yourself politely if you have never met them before. A simple example: Hi Qing, I noticed your name on your badge. I have read your work on xyz, and wanted to introduce myself. <Plug your introduction>
[III] Grow your network
To really make strides in growing your network, think about these key steps:
- Looking for opportunities: keep your eye open for potential opportunities. Visiting the industry booths, networking with folks during socials, attending the posters, everything is an opportunity to meet and connect with other researchers in the field.
- Finding good mentors: can be a game-changer. Look for experienced folks who can offer advice and share their wisdom. Don't be shy about connecting with others in the field too! Potential collaborators can bring fresh perspectives and help you tackle bigger challenges or validate your ongoing journey as well. Seek and attend mentoring sessions to find lifelong mentors.
- Connecting with potential collaborators: explore meeting authors whose papers struck a chord with you. It might even be that you are building on another groups' work, interacting with them might open doors to your next research project together with them.
- Affinity groups: explore affinity groups or online communities such as Cohere Discord. Finding your people – those who share your interests and understand the journey – can provide incredible support and a sense of belonging as you navigate this fascinating field. A few examples include Women in Machine Learning, Women in Machine Learning-Theory Black in AI, Latin X in AI, Muslim in ML Community, etc.
[IV] Look beyond the horizon - cultivating a forward thinker
Developing a long-term, impactful research agenda in the field is valuable and requires more than following the latest trendy topic. Published work is by definition a lagging indicator. To truly innovate, cultivate the ability to have a broader vision, anticipate future trends, and connect your work to developments in other fields goes a long way.
Machine learning receives stimulus from other fields. Attending a conference can shape this forward thinking perspective in many ways:
- New mathematical tools in maths and stats often inform the analysis of ML problems. For e.g. development of optimal transport theory has found significant applications in generative modelling.
- Up-and-Coming Software System developments such as specialized hardware (GPUs, TPUs, etc.), or new programming paradigms could impact both day-to-day deployment of experiments, the infrastructure and corresponding findings.
- Meaningful datasets and benchmarks from across the field can influence your work. Critical thinking about the limitations of existing benchmarks can lead to new evaluation metrics and datasets that better reflect real-world challenges.
- Finding your niche in foundational research that often takes time and persistence, comes with curiosity and a long-term perspective. Chatting with researchers from different sub- and across- fields during the conference can greatly shape your thinking in honing your niche.
Practical actionable tips on what to do
Before the conference
- You have a website, right? RIGHT? You must have one if not. There are many quick ways to get one, here are a few templates to get you started.
- Prepare "elevator pitches".
- An introductory, 60 second, pitch about yourself. Note that this pitch is simply your introduction, who you are, what you do, etc.
- For discussing your research, three different 1 minute stories:
- What is the problem/question your research tries to answer?
- How your approach provides a solution to the question
- Where does your research question (and your solution) apply in the real world? Why should anyone care?
- Make sure your presentation or poster is ready before you travel.
- Don't waste the precious time at the conference stuck in your room preparing one.
- Note many conferences offer to print, if you follow their emails and instructions in due time.
- Email and schedule meetings in advance
- If you know authors that are presenting and find their work interesting, do not hesitate to reach out and schedule coffee-chats.
- If you read a prof's work and would like to collaborate with them, do not hesitate to reach out, email them.
- Anyone at all you want to talk to, they may be busy and not have to meet with you, but emailing them is your best shot at increasing the chances of this meeting to be non-zero.
- Looking for internships? Another reason to schedule these meetings with potential recruiters ahead of time. Sometimes meeting folks who might not be hiring themselves can also be very fruitful.
- Pencil all events in your calendar
- Once your plans to attend solidify, and the program schedule is out, it can be very productive to slot talks, posters, and socials you are attending in your calendar.
- This also opens up clarity of view when you are available to schedule the aforementioned chats with individuals.
During the conference
- Organize yourself
- First things first, register to get your badge, don't lose your badge.
- Secure your valuables like passports in the hotel safe if you can, and be safe.
- Try to have your pre-worked calendar on your phone so you can follow along smoothly where you need to be when.
- Be selective, intentional, and productive
- As tempting as it might be, do not try to attend everything. Commit to whole sessions, instead of jumping up in the middle of sessions. It is more important to talk to people rather than listening to all talks.
- Take notes. E.g. interesting posters, references to interesting work published elsewhere. David Abel's notes are an excellent examples of conference notes.
- Be mindful of people's time: if you committed to chat with someone, show up on time, or communicate.
- Follow the flow.
- Allow for serendipity.
- Pace yourself. You are probably jet lagged, and learning a lot. Make sure you are still as energetic and not getting sick by the second day.
- Explore the host city:
- Respect and learn about the local culture. Conferences are usually at interesting places in the world.
- Check out some must-try local food and use this time to meet old friends and make new ones.
After the Conference
- Follow up
- Within 24 hours of any conversation, write a message (social media including email), and potentially connect with them on Linkedin or other social media platforms.
- Point to specific URLs for any resources (papers, software, data) that you have mentioned to people. Not just your own work, but also other work that you found interesting.
- Reflect
- On your notes, on your research, and next steps
- On the new connections you have made
- On the field as a whole
- Share your insights with your colleagues at home.
- Organize key paper summaries for your lab
- Report any interesting work that stood out for you
- Share your notes with the broader community
Small vs large (> 500 attendees) conferences
Some extra tips for when the event has a lot of participants
- Additionally use online tools to organise to meet in person. Many conferences have conference chats, such as Whova.
- Most large venues have a whole day dedicated for affinity group workshops.
- Remember that some participants (especially people who were adults before smart phones) are not permanently online. And most people would not have notifications turned on.
- As a early career researcher, if you recognize someone you have met before, but they did not, do not take this personally. It's them and not you! Think kindly of the busy-ness of extremely large conferences which transfers to the busy-ness of researchers as well.
- Industry sponsors often host social/parties, which are often invite only. Go to their booth, ask for an invitation. These events are a good place to find opportunities or simply network.
Top 10 Reading Recommendations
If you are new in ML, in addition to this blog post, you might potentially find some of the following resources useful. We have curated top 10 articles, blogs, or venues.
- How to Be a Successful PhD Student in Computer Science by Mark Dredze and Hanna M. Wallach.
- Richard Hamming’s classic lecture on You and Your Research
- Researcher evaluation by UK Royal Society
- Dirk Hovy on How to navigate post PhD academia
- Devi Parikh's blog on time management: Calendar. Not to-do lists
- Ben Recht on "What if, hear me out, we are writing too many papers?" Too Much Information.
- Reviewing guide by NeurIPS
- Devi Parikh on How we write rebuttals.
- Lilian Weng documenting learning notes in her blog since 2017.
- New in ML Recurring Workshops.
Authors
Khimya Khetarpal is a Research Scientist at Google Deepmind and an Affiliate Faculty Member at Mila.
Cheng Soon Ong is an Associate Science Director at Data61, CSIRO and an Adjunct Associate Professor at the Australian National University.
To cite this blog
Â
@article{khetarpalongnewinml,
title={New in ML: A Guide for Navigating the ML Conference Scene},
author={Khetarpal, Khimya and Ong, Cheng Soon},
year={2025}
}
New in ML: A Guide for Navigating the ML Conference Scene
🔖 Tldr: Together, Khimya Khetarpal and Cheng Soon Ong go deeper in this blog on building a resource for new and early career researchers when attending a Machine Learning (ML) Conference.
Intended Audience
This blog post is targeted at people who are (almost) first time participants, whom we are calling "New in ML". It is most suited for folks who might fall in one of the following categories:
New In ML: Attending your first in-person Machine Learning conference?
Goals of conference participation
Considering you are New in ML, to make the most of this time when attending in person, we list a few of our personal recommendations as potential goals of your attendence.
[I] A core component: research
A core and large part of attending a ML conference is about research. Centered around research, the attendees have goals ranging from presenting their own research, to attending talks, to meeting researchers who might have overlapping research interests, to finding researchers with similar backgrounds or aspiring trajectory paths, etc.
Skimming through posters, accepted papers beforehand is useful to know what topics the conference broadly casts a net over. If you find related and relevant work to your current projects, this might look scary at first, and give a sense of insecurity. However, this implies more than one lab is investigating this particular problem, it is indeed an open problem worthy of studying, and that you have competition as well, which is healthy and adds to your research.
[II] Advertise your research, advertise yourself
If you are presenting a paper (either oral or poster), while it is true that the work will speak for itself, we highly encourage you to advertise your work! You have to be your own hype person! If your advisor or any collaborator is attending you could also encourage them to advertise the work. You are also advertising yourself (both your skills and also your ability to work in a team) as an interesting colleague or collaborator. You may be representing the place you come from, and the place you are currently working at.
But how? There are several ways to do this.
[III] Grow your network
To really make strides in growing your network, think about these key steps:
[IV] Look beyond the horizon - cultivating a forward thinker
Developing a long-term, impactful research agenda in the field is valuable and requires more than following the latest trendy topic. Published work is by definition a lagging indicator. To truly innovate, cultivate the ability to have a broader vision, anticipate future trends, and connect your work to developments in other fields goes a long way.
Machine learning receives stimulus from other fields. Attending a conference can shape this forward thinking perspective in many ways:
Practical actionable tips on what to do
Before the conference
During the conference
After the Conference
Small vs large (> 500 attendees) conferences
Some extra tips for when the event has a lot of participants
Top 10 Reading Recommendations
If you are new in ML, in addition to this blog post, you might potentially find some of the following resources useful. We have curated top 10 articles, blogs, or venues.
Authors
Khimya Khetarpal is a Research Scientist at Google Deepmind and an Affiliate Faculty Member at Mila.
Cheng Soon Ong is an Associate Science Director at Data61, CSIRO and an Adjunct Associate Professor at the Australian National University.
To cite this blog