Queering Machine Learning

Permit me to introduce myself briefly. My name is Shakir Mohamed; I use the pronouns he/him/they. I am exceptionally grateful that the journey of my life so far has led me in some way to the point where I can be engaged with you in this method. In truth, that is what Shakir means-its an Arabic word for a grateful individual. I am a proud South African and was born during the late apartheid age, and also born into the religious beliefs and traditions of Islam. Much of my memory is of the city of Johannesburg, where I was born and bred and lived until I delegated do a PhD in Machine Learning at the University of Cambridge. So these are a few of the prerequisites for much of my journey towards self-esteem, and ultimately self-love– including all the typical suspects of crises of intellect and faith and body and sexuality and racialisation and culture and location. Today, I am gladly silently confidently queer. These earlier crises do still resurface, however in different methods, and still being not sure, each day I continue to explore my location and voice in this world. So all this is to draw you closer to me, and with that bond start a conversation about our duties and roles as queer researchers and residents.

What I hope you will see when we observe our work using this hierarchy, is that the questions we ask are for the many part epistemic questions. They are concerns of knowledge that we think are very important in our field and that drives what we deal with and think about cutting-edge. This hierarchy shows the epistemic values we utilize in our research – worths like reproducibility, the role of depth and attention, significance of calibration and unpredictability, a concentrate on reinforcement knowing in establishing smart systems.

I postured that opening question: why do we organise, for 2 reasons. To see that the queering of device learning offers an unique view on technical systems and their style and release, that is our strength. And secondly, to join with you to create a neighborhood of support and belonging, where security and isolation are not to be feared, however used to create a more important and accountable field of artificial intelligence.

Ive entitled this talk Queering Machine Learning, which is theme I wish to explore with you today. To get us began I d like to provoke your thinking with a question..

Based on what youve simply made a note of, I hope that what I will speak about for the next 15mins approximately will have some resonance. And I hope we can utilize your reactions and responses as a style for follow-up conversations throughout the week..

One of my responses to this concern is to describe machine learning as a procedure of producing pathways from concepts to products. What I hope you will see when we observe our work using this hierarchy, is that the questions we ask are for the most part epistemic questions. Other examples consist of: the natural language tools that categorize queer vernacular as poisonous in online forums, trans bodies that are methodically singled out by airport screening systems developed only for a gender-binary world, and in questions of algorithmic fairness that are presently established while limited to binaries of gender and race. The queering of the technical landscape of machine learning is one important response to that concern I posed to you at the very beginning.

At the bottom are the principles upon which our field is built. Those concerns of probability theory, mathematics, asymptotics, neuroscience and cognitive science, and lots of other topics.
Building on that we have informative questions. This is where questions of forecast, uncertainty, and discovering with depth emerge.
This then leads us to more intricate questions of reasoning, whether long-lasting preparation, world-simulation, or description.
And at the top of this hierarchy are applications, in healthcare, environment, and numerous others where our company believe artificial intelligence can have energy, and that inspires a lot of our work.

Why are we here at QAI? Or asked in another method: Why do we gather and arrange ourselves in this way? Why do we arrange?

Let me now pose an extremely various concern to you. We are together at a virtual conference on maker learning. So, What is machine knowing?

We can currently see this exact same kind of failure starting to manifest in artificial intelligence as well, with examples are plentiful. There is the notorious case of the image classifier established to categorise people as gay-or-not from their faces alone. What makes that scene even worse, is that the scientists included stayed relatively oblivious to the possibility that their work was extremely suspect, which they had actually fallen into the trap of essentialising gay people..

Since of the dominating cultural attitude of the time– i.e. Victorian contextual worths– this work stopped working to recognise the same-sex behaviour observable in the animal kingdom and the long recognized variety in human desires, so rather, was just able to conceive of a function for men in relation females, and became a manifesto for heteronormativity and gender hierarchy. This is a failure and the consequence of what occurs when we stop working to question the contextual values embedded within our sciences.

Motivated by a number of the conceptual hierarchies we utilize in computer system science, hierarchies like the classical cognitive architecture, Marrs levels of analysis, or Suns phenomenological levels, I often describe maker knowings paths utilizing my own 4 layer hierarchy..

Some books and recommendations.

Her message is as relevant today as it was when she first wrote this poem. Her advocacy, nerve, and composing changed the world, and reminds us that looking for other futures is what we, who live at the coastline, can and need to continue to do.

As a last response, I think Ill rather acknowledge something more easy: that there is joy in being together. Pleasure in the possibilities of technology and what it can do in the world is what brought us to maker knowing, and delight is what we see, whether personally or practically, when we satisfy each other at this Queer in AI and all the others that have and will come. Happiness can create a world of diversity and distinction that we can be happy of, and happily defend.

This kind of belonging is important. And of the society we live in since it is just with this type of safety that we can be more vital of the work we do. With this kind of belonging, we permit ourselves to be one step removed, to put in the time to be in solitude, to not take the stories and messages we inform of our work and schools and offices as self-evident truths, but to look at them for what they are, and to take the action, utilizing Audre Lourdes words, that seeks a now that can reproduce Futures, where all of us belong.

The paradox is that isolation and belonging, go together. Thats because real belonging, really genuine belonging, always allows people to be alone– the much better word to use here might be privacy. It is in producing this kind of belonging where many communities stop working. This is also a deep point about addition that is often ignored in a lot of the deal with D&I- Diversity and Inclusion. When we truly and genuinely belong, no matter how divergent my opinion might be, how crucial a viewpoint I might take, or in the face of the mistakes I may make, I will know that I still belong, therefore do you..

What an experience we are all having in taping these videos! Of all the videos you could be viewing, Thank you for enjoying this one, and being here – Im honoured to be given the present of your time. And naturally, a big thank you to the organisers of the ICML2020 Queer In AI workshop for this opportunity.

Perhaps this is a 2nd response to my opening concern. By arranging in artificial intelligence, and by queering device knowing, we construct collective community and collective strength that makes it possible for belonging and solitude and solitude to co-exist and strengthen each other for the advantage of our field.

Specifically for queer neighborhoods, where so many understand what it indicates not to belong. Not belonging in school or university is hard. And how much harder it is, to go house and still not belong.

Keep in mind the work of Edward Carpenter, the late Victorian intellectual and activist, who utilized his queerness to work versus imperialism and expose it for all its evils. Specifically in this time, where we are rising up worldwide in assistance of #BlackLivesMatter and against the ills of bigotry that is a direct repercussion of slavery and empire, we discover that queers and queerness, regardless of all the difficulties they themselves faced, were allied to the cause of racial justice and equality from the earliest times.

The worldwide coronavirus pandemic, our resistance to racial oppression, and our concern for queer neighborhoods everywhere advises us that to demand belonging, community, and human connection, is as crucial today, and in our field, as it has ever been. Due to the fact that there is delight in our belonging.

There is another, maybe more vital, action to my opening concern of why we gather and organise. For much of us, I suspect the response to that question was instinctive: to belong. If we think about all our customs and narrative histories, of the excellent works of literature that record life and culture, and consider all our psychological responses in times of happiness or crisis, something stands apart, and that is a consistent search for belonging. In being together, there is a connectedness and a collective self-confidence that we develop. And In being together, we make it possible to develop a brand-new language with which to name the coming technological challenges I mentioned earlier.

I am incredibly humbled to have actually been able to give a brief talk at the Queer in AI workshop at ICML2020. This is the text of the talk. Enjoy the video. Browse the slides.

That Black lesbian mother warrior poet Audre Lorde, in the opening lines of her poem A Litany for Survival, wrote:.

In my own operate in variational reasoning I have begun right at the bottom as we were thinking of concerns of approximate Bayesian inference and the differences between prescribed and implicit generative designs. And in other cases, I have actually begun right at the top, as we took on the obstacle of developing systems to support the early forecast of organ damage in hospitals utilizing electronic health records. Each people will begin at various places in this hierarchy, and at various levels, and then access different parts of it in different methods. This chance we have to switch techniques and tools and perspectives is, for me, why artificial intelligence is so amazing a field to work in..

We are what the atmosphere is, transparent, receptive, pervious, invulnerable,.
We are snow, rain, cold, darkness, we are each item and influence of the globe,.
We have circled around and circled around till we have actually arrived house again, we two,.
We have actually voided all but freedom and all however our own pleasure.

The queer activist Simon Nkoli declared at the dawn of South Africas democracy:.

Once again you can indulge me by stopping briefly the video to jot down a few of your own ideas. And I hope youll use the chat to share a few of your answers to this question with others..

Other examples include: the natural language tools that classify queer vernacular as poisonous in online forums, trans bodies that are methodically singled out by airport screening systems developed only for a gender-binary world, and in questions of algorithmic fairness that are currently established while restricted to binaries of gender and race. And these are among numerous other emerging concerns varying from the look for gay genes in genomics information, the online tracking and doxxing with the intent to out, daunt or discredit, all the method to discrimination in working with and the provision of social and health care services. The queering of the technical landscape of artificial intelligence is one essential action to that question I presented to you at the very start.


For many of us, I think the answer to that concern was instinctive: to belong.

Thank you.

Ive asked this concern often times, and it is a surprisingly difficult one to respond to. I suppose this reflects simply how diverse machine learning is as a field. Among my answers to this question is to explain machine learning as a process of producing pathways from principles to products. This is an effort to identify that our field is both a clinical field, dedicated to advancing the theoretical and statistical basis of learning from information; along with an engineering field, working towards building systems that can be released in the real-world which adapt to the needs of noisy, complex, impactful issues..

Interpret this question in as narrow or broad a method as you like. Since you are enjoying a recording, indulge me by pressing the time out button to take 2 minutes to compose down any responses or keywords you have to this question.

” In South Africa, I am oppressed since I am a black guy and I am oppressed due to the fact that I am a gay man. When I fight for my freedom I should fight against both injustices”.
Simon Nkoli, Joburg Pride 1990.

Simon died of HIV/AIDS in 1998, and thinking about him reminds us that by queering the values and systems we exist in, we can change the shape of whole democracies. Here, Im believing of this impact on rights enshrined in the South African constitution. Im also thinking of the lots of incredible people, like Binyavanga Wainaina from Kenya, who regretfully likewise died last year, Jason Jones from Trinidad who is improving the rights of millions across the Caribbean, the Tongzhi in China, and so numerous others queering for their rights.

This hierarchy stops working to identify that our field is not formed solely by epistemic worths and questions.

Queerness in science might best thought of as a verb: to queer. When we queer something, we utilize the human experience of desire and identity and relations to reveal that ideas and tools that we consider granted, may be the reverse of what they appear to be. We are able to expose that our work and techniques and thinking already has even more range than what is visible or mainstream. This is a queering of device learning, and a powerful tool of self-refelection; a technique to artificial intelligence research that is more accountable and vital; a tool offered not just to queer scientists, however to everyone.

Ill end here by reading the last few lines of a poem by Walt Whitman, We two, how long we were fooled.

To find out more on this theme, checked out a short piece on Queer Exceptionalism in Science, and on race in Racialised Lives and the Life Beyond, or thoughts on A New Consciousness of Inclusion in Machine Learning.

I believe we have adequate proof in history of what is possible when the queering of our world is taken seriously.

Leave a Reply

Your email address will not be published. Required fields are marked *