Raymond Ng kicks off the @UBCDSI Data Science Symposium, with @SohrabShah as the first speaker #ubccs50 
Ovarian cancer investigated through clonal maps to see differential spread and virulence across locations within the body @SohrabShah #ubccs50
As a visualization person I'm struck by the number and diversity of the visual representations used in this talk... @SohrabShah #ubccs50
... and I particularly like the e-scape work for cancer visualization with Maia Smith and @cydneybn, more at shahlab.ca/papers/28557980 @SohrabShah #ubccs50
Wrap-up with lessons learned from high dimensional and genome and computational analysis in cancer. @SohrabShah #ubccs50
is heading to Sloan Ketterling and he's hiring! #ubccs50
Continuing the theme of data science and cancer with Raj Chari of the Frederick Natl Lab for Cancer Research, on emerging opportunities for large scale data analysis in eukaryotic genome engineering #ubccs50
From CRISPR to microfluidics... #ubccs50
Now up: Heidi Lam on Building Tools to Support Analysis, handling large and multidimensional datasets #ubccs50
Scenario: data preparation. What could go wrong? Pretty much everything, NULLs galore... #ubccs50
Identifying the subsets of the data that matter for a particular analysis question, tool from Google (shown here with synthetic data) #ubccs50
Beware confirmation bias...#ubccs50
Identifying unusual items - aggregates are not enough, need to see the details. Example of tuning search results with SessionViewer tool used inside Google #ubccs50
What can go wrong? Being shunted into Google Scholar to read concurrent programming academic papers when you want to know color of umbrella in a Seurat painting... But don't try this at home, they've fixed it now! #ubccs50
Summary from Heidi Lam: understand situation, understand aggregate from data items, identify differences between subsets, identify unusual items based on distributions. #ubccs50
starts off the AI Changes Everything keynote and panel downtown #ubccs50
Keynote from @erichorvitz #ubccs50. Starts with Pillars of AI, gives many examples of industrial applications including handwriting recognition on snailmail envelopes, then argues for how it will transform science.
Aspiration 1. Gaining human level intelligence, beyond narrow wedges of competency. Pursue understanding of mysteries of human intellect. @erichorvitz #ubccs50
Integrative AI - humans are not just one thing, machines can be built from many pieces too... @erichorvitz #ubccs50
Complementarity to extend beyond the human and catch errors, hoping to avoid cognitive bias and blind spots. Diagnosis example: Expert alone error rate of 4% better than AI alone, but combination of both gets you below .5%. @erichorvitz #ubccs50
Nice examples of complementing and scaffolding decision making by humans, rather than attempting to completely replace them. @erichorvitz #ubccs50
Mixed initiative example of suturing example handoff between human and machine. Much wincing at example video before hearing that it's cookie dough not human skin, whew @erichorvitz #ubccs50
Third aspiration: Leveraging AI on societal challenges. @erichorvitz #ubccs50
Panel Q&A: @gail_murphy points out the complementarity of new AI for Decision Making and Designing for People (@dfp_ubc) initiatives #ubccs50
Panel Q&A: regulation could play helpful role, consider successes of NTSB for making air travel so remarkably safe. @erichorvitz #ubccs50
The "AI winter" was the most productive time for the field.... But still, hype is high now. Duty of technical people to inform and demystify for non-technical leadership of companies @erichorvitz #ubccs50
Mark Schmidt: one big difference this time around is the amount of money on the table, massive investment. Arguably unsustainable level of new company formation. But like last time, there will be jobs for good people no matter what. #ubccs50
Q: how to weight rare AI mistakes vs pervasive non-AI mistakes of humans? @erichorvitz: High profile AI mistake could have chilling effect for years. Mark Schmidt: I'll be really disappointed if we don't do better than humans #ubccs50
: hard to predict impact beforehand. Consider: What did electricity do to the economy? #ubccs50
: Notion of a job is not that old in human history. Maybe we will reorganize in a massive disruption... #ubccs50
Kevin Murphy A: Childlike stuff might be interesting research but "first do no harm" principle applies for applications in world. Need to start with a minimum level of competence and improve from there. Can have algorithmic guarantees to stay above initial line. #ubccs50
Unravel the magic - @gail_murphy. Adversarial examples can show the complexity - Kevin Murphy. Wish the term coined in 1950s was computational intelligence rather than artificial intelligence, which appears to exclude humans - @erichorvitz. #ubccs50
HCI Symposim about to start! #ubccs50
More on DFP at dfp.ubc.ca #ubccs50
Karon MacLean on the grand challenges of haptics that her SPIN lab addresses cs.ubc.ca/labs/spin/ #ubccs50
Joanna McGrenere on one of the projects from her group on digital preservation tendencies - the continuum between hoarding and minimalism (recent best paper award!). More at cs.ubc.ca/news/2018/05/c… #ubccs50
Our new faculty member Dongwook Yoon on multi modal interaction #ubccs50
of UW with a great overview of her research on AR for accessibility #ubccs50
from MSR -- the third PhD to graduate from UBC HCI -- on engaging with technology through play, transcending distance, how video captures emotion more easily than text, integrating the physical with the virtual. Technology probes to explore what's possible #ubccs50
Melanie Tory @vizstudylady from @tableau Research on Conversational Analytics - how to bring fluidity of human back and forth talking to interacting with data. Eviza and Evizeon demos, more at research.tableau.com/paper/applying… #ubccs50
: I love this comic as the mother of a teenager. Me: wow, what an evocative example of prgamatics (branch of linguistics concerned with intent)! What does "actually yes" mean? "Actually no"... #ubccs50
Opportunity - 35% of utterances referred to previous in one study. Challenges - if you get the pragmatics right then users expect it to be even smarter! @vizstudylady #ubccs50
Brian Gleeson of Apple in the HCI/UX Industry Panel, discusses many contrasts.
great work enabled by:
acdemia: more autonomy, vs
industry: more resources
#ubccs50
great work enabled by:
acdemia: more autonomy, vs
industry: more resources
#ubccs50
More from Brian Gleeson on academia vs industry.
spend time on: writing grants vs attending meetings
struggle with: impact vs secrecy
#ubccs50
spend time on: writing grants vs attending meetings
struggle with: impact vs secrecy
#ubccs50
Brian Gleeson - constraints: sometimes ignore, sometimes embrace... the ideal, provacative vs the practical, low risk. #ubccs50
of Samsung: how does UX research work in real world vs what
he learned in grad school? #ubccs50
he learned in grad school? #ubccs50
research team needs two things: center of excellence, and be perfectly
positioned. grad school really helped with the excellence, but needed
to learn a lot about positioing.
what is a ux researcher's role? who a ux reseacher works with? how to
bring value? @rockleung #ubccs50
positioned. grad school really helped with the excellence, but needed
to learn a lot about positioing.
what is a ux researcher's role? who a ux reseacher works with? how to
bring value? @rockleung #ubccs50
UX researcher's roles: scientist, facilitator, auditor. he gravitates
towards facilitator, but might need to wear another of these hats at
need.
- @rockleung #ubccs50
towards facilitator, but might need to wear another of these hats at
need.
- @rockleung #ubccs50
who does a ux researcher work with? humbling realization: you can't go it alone.
- @rockleung #ubccs50
- @rockleung #ubccs50
concludes with a plug for radical research summit! fri sep 28, 2018 here in vancouver, call for speakers is out now, tickets on sale later
radicalresearchsummit.com
#ubccs50
radicalresearchsummit.com
#ubccs50
Next panelist: What Barry Po learned doing ux research in the real world.
Anecdote: doing a careful user study in the field. CEO walked in, watched one user, said "I'm convinced", and walked out.
not everybody cares about sample size...
#ubccs50
Anecdote: doing a careful user study in the field. CEO walked in, watched one user, said "I'm convinced", and walked out.
not everybody cares about sample size...
#ubccs50
Barry:
1) focus on outcomes instead of outputs!
limited energy, so be clear on how you'll have an impact
2) rally people to your cause by finding ways to help them!
you'll always be outnumbered, connect with things other people care about
3) pick your battles!
#ubccs50
1) focus on outcomes instead of outputs!
limited energy, so be clear on how you'll have an impact
2) rally people to your cause by finding ways to help them!
you'll always be outnumbered, connect with things other people care about
3) pick your battles!
#ubccs50
Next panelist: Brad Swerdfeger from Shopify. Building the Latticework. How I stumbled into being successful in my career. Taking the 80% luck and turning that into opportunities.
#ubccs50
#ubccs50
Brad: starting from cogsci at Queens and then CS at UBC. Need to use hard skills to get in the door, then use lattice of mental models to succeed #ubccs50
Industry HCI/UX Panel final Q&A tidbits
barry: universities can teach empathy, not just for users but for other stakeholders too.
brad: delegation is your best friend. you may not have time but you do have the resources
#ubccs50
barry: universities can teach empathy, not just for users but for other stakeholders too.
brad: delegation is your best friend. you may not have time but you do have the resources
#ubccs50
Now up: MUX lab tour in X508/X521 across the street! One of these rooms is brand new. We'll have a lot of videos playing about our latest work, it's a five-floor detour (optional of course!) on your way to lunch in ESB #ubccs50
The CS++: Towards the Next 50! Panel begins. David Cheriton delighted that Margo Seltzer will be the Cheriton Chair. They know each other in part through high performance transaction processing - since low performance transactions are just plain irritating! #ubccs50
actively thinking about transformation of CS at the 50 year mark. A glutton for punishment: wasn't enough to do PhD, then to do a startup -- now he's writing a book #ubccs50
Chris Thachuk argues for the integration of matter and computation #ubccs50
thinks the coding baboons will be the last to go extinct (continuing with Cheriton's metaphor) #ubccs50
So many ideas getting tossed around in this great panel that I can't nearly keep up. Apocalyptic examples, flying baboons, are humans or computers really flying the plane, ... #ubccs50
Seltzer: before we think about CS literacy we have to worry about much more basic science literacy - once you lose that, very bad things happen #ubccs50
Seltzer: I work on scientific reproducibility and it's still a fantasy. Only 15% of scripts in Harvard code/data repo could be executed today, and 30% of them could not have run ever since were missing crucial parts when uploaded. #ubccs50
Fun seeing historical pictures and hearing stories at dinner! Research awards to Monica Lam and David Lowe. Entrepreneurship awards to @mik_kersten of @Tasktop, Point Grey Research, and Murray Goldberg of many startups including WebCT. #ubccs50
And so lovely to see the founders awards go to Charlotte Froese Fischer who taught the first classes in CS at UBC and Richard Rosenberg who started the department! #ubccs50
