Conference summaries, Oct 2017 part 2.
Continuation from Conference summaries, Oct 2017 part 1. — as a summary of international conferences that I participated in during 2017.
London. A Coruña. Madrid. Singapore. Oahu. Got to wander far from mainland US in 2017, mostly for professional conferences about Big Data, Data Science, Artificial Intelligence, etc. Along the way, I encountered wonderful intellects, vistas, trees, beaches, dogs, new friends, music, food, and — rather importantly—cider. Here’s the tale…
Just slightly before Strata UK last May, someone invited me to register for a performance in London called “Partnerships”, about music composed by AI, as well as musicians performing with AI accompaniment. I thought this had been sent as part of Strata, although Gina Blaber and others folks on our Conferences team swear that wasn’t so. Perhaps some bizarre quantum entanglement crafted a note for my email inbox. Or ghosts. Anywho, I got registered online, then found my way through the twisty cow trails from Victoria Dock to Stepney High.
The performance was a collaboration between Bob Sturm at University of London and Oden Ben-Tal at Kingston University, and friends. Their research in AI-augmented music is largely based on a recurrent neural network project that’s available as open source on GitHub: folk-rnn.
In other words, feed a bunch of musical scores — more than 23,000 songs were used for the training set in this performance — into an RNN, then let it generate new musical scores to perform. They’d focused on Celtic folk music, which has a structure that lends itself to working with the data representation they’re using for deep learning.
To wit, I found the melodies pleasing, emotive, eerie. Hair-raising at times. It was clearly music, folk music, specifically familiar as Celtic folk music, albeit ineffably different. I kept being reminded of the AI in William Gibson’s novel Count Zero, who spun out faux Joseph Cornell boxes, conveying to characters in the know “the suggestion that any environment might be unreal.” One of the performers, harpist Úna Monaghan from Northern Ireland, mentioned what she calls “QBW” — about parts where the music has quirks, but it works. Her performance was fantastic!
Bob Sturm performed The Humours of Time Pidgeon, March to the Mainframe, and Interlude. The latter was among my favorites, in keeping with the Celtic folk theme. Ensemble X.Y performed Bastard Tunes, generated by folk-rnn then arranged by Oden Ben-Tal.
Other related stories about AI and music:
Departing London, the taxi barely made it to Heathrow on time after we got stuck in the midst of a bomb-scare evacuation — literally, stuck in gridlock while police were making people run away from our location — near Tower Bridge. Nonetheless, I found my way to beautiful A Coruña on the coast of Galicia.
Big Data Coruña 2017 was organized by the incomparable Amparo Alonso-Betanzos, who leads the Laboratorio de Investigación e Desenvolvemento en Intelixencia Artificial at University de A Coruña. I gave two talks, including Human in the loop, though my favorite talk was Block chain application and vision in the banking industry by Michael Bassios at Data Spartan. Watch for more from Michael, he’s brilliant and an excellent speaker. Looking forward to BDC 2018!
Imagine the area near Santa Cruz, if it were located in Spain instead of California. A Coruña features a temperate maritime climate, an excellent university, incredible food, world class surfing competitions, and rows of Albariño vineyards climbing up verdant coastal hills. I had a childhood fascination about the historical period of Sub-Roman-Britain, which I recently learned Robert Plant shared — speaking of passion for British folk music and cider, which are also among his hallmarks. Britons had well-established maritime trade routes between Wales/Cornwall, Breton, Galicia (draw a line on the map). Those who fled the incoming Saxon overlords sailed mostly to Brittany or Britonia, and the latter is now Galicia.
Amparo took me to see the local seafood markets on the weekend following Big Data Coruña, a rare treat. While we were at a stall buying local cheeses, a group of bagpipe players marched loudly down the aisle. You’ll encounter people with bright red hair, sometimes wearing kilts, playing bagpipes — as well as jewelry that looks vaguely Welsh, linguistic elements where many more geological features are feminine than in Spanish, and other vestiges of Celtic culture. Plus, of course, excellent alcohol :)
After the markets, we drove down to Santiago de Compestela, where Amaparo did her PhD work. University of Santiago de Compostela was founded at an auspicious time for the Old World: during the second voyage of Columbus. Renaissance Europe, and in particular this region of Europe, was abuzz with fantastic stories from the “newly discovered” Western Lands —although the Vikings and Welsh have claims to arriving earlier, as do the Chinese.
Following our incredible meal of local wines, cheeses, and charcuterie in a dark and ancient Galician cellar, I found myself staring wistfully at what is now the Computer Science building. As a visitor from the New World — especially coming from Silicon Valley, here to discuss some of the latest technologies— I had a sense of glimpsing back 500 years, to a time when this university had been among the most advanced technological centers of the Old World. Façades of science buildings, which in earlier centuries featured Roman Catholic triptychs — pictorial stories about the lives of saints — have since replaced their religious parables with histories of breakthrough research efforts by famous Nobel Laureates. In another 500 years, let’s say in the year 2512, will some visiting scientist from a thriving settlement on Mars — upon returning to tour the “old country” in Silicon Valley — smile at the terra cotta buildings and courtyards of Stanford University with a similar sense of nostalgia?
Rolling the clock forward a few months, I returned to EU to lecture (briefly) in London, then continued on to Big Data Spain in Madrid.
Aside from a quick pint in Gatwick, I didn’t get a chance to visit my favorite cider venues in London, especially The Williams Ale & Cider House. Nor a chance to chat with good friends at Shoreditch start-ups. Much to do in Old Blighty on my next trek through. Meanwhile, I’ve heard that The Stable is the recommended place to go for taps — they’ve even got cider sommeliers! Let’s gather for a data drink-up at one of their locations during Strata UK 2018.
This 6th edition of Big Data Spain was packed full of interesting people, enlightening ideas, compelling content — not to mention, outstanding cuisine! Their stats speak for themselves: 74 speakers in 4 tracks, over 2 days; 8 half-day training courses (6 sold out, 2 cancelled); 1200 attendees; 30 sponsors with booths. Outstanding work by Rubén Martínez, Lidia Fernandez, et al.
I arrived early to teach a course, “Intro to NLP + AI”, with Daniel Vila Suero, co-founder of Recognai in Madrid and an expert in NLP and AI. Our course was based on spaCy and PyTorch, plus some of Daniel’s fantastic integration of AllenNLP and TensorBoard for working with PyTorch.
Daniel’s doctoral dissertation in ontology engineering at UPM was adopted by the national library of Spain as the ontology used to organize their content. People flew in from other parts of Europe to attend our course, and I’m pretty sure that was to catch Daniel :)
Paradigma Digital, the organizers for Big Data Spain, know how to treat their keynote speakers well. We went out to a restaurant in Madrid, Colonial Norte, where the sommelier had prepared a special workshop. She guided us through a tour of tasting Spanish wines, along with how to pair them with tapas. Of course, one must learn through ample hands-on experience, which is the best method. I struggled to avoid appearing très-Mhurican, sounding like a gushing fanboi for Godello — no matter how damn good that wine is! Amparo assures me this is simply because I haven’t yet become sufficiently acquainted with their Albariños, etc.
My favorite talks included: Irene Gonzálvez from Spotify presenting Big Data, Big Quality? and Verónica Bolón-Canedo from U de À Coruña presenting Feature selection for Big Data: advances and challenges. Irene’s talk was about Data Quality practices at scale at Spotify — practical, thought-provoking advice. Verónica’s talk was about her research along with Amparo, which completely changed my thinking regarding how to handle missing values, etc. Compelling work, highly recommended.
I should mention that Holden Karau was arguably the star of #BDS17, which she quite well deserved. That audience is eager to hear about Apache Spark!
My keynote talk was Human-in-the-loop: a design pattern for managing teams which leverage ML:
Also had an interview by María Cedrón, «Estoy a favor de una renta cubierta por las tasas de las empresas» about the implications of AI vis-a-vis jobs, universal basic income, etc. That ran in La Voz de Galicia alongside her (IMO, more impressive) interview with Amparo, «Hay que legislar para ver cómo hacer esta transición imparable».
No trip to Madrid can be considered complete without mention of sidra.
A couple of my friends, Ⓜünchausen and Marco Laucelli, helped school me about the proper sidra options and venues, particularly those from Asturias. You can read our thread on Twitter, but to summarize:
- sidreria.com is a portal about Asturian sidra and where to find it — e.g., best venues in Madrid and other cities in Spain
- Carlos Tartiere is highly recommended, especially for their Sidra Cortina Coro
- Zerain is another recommended sidreria, in the Basque tradition — Paradigma had taken us there
- friends at Paradigma slipped me a bottle of Sidra Trabanco Cosecha Propia to carry home, which was delish!
- Next time through, I’d love to check out Casa de Asturias
Meanwhile, two friends who live in Madrid — Sandra Cobos and Lisa Lange — made sure I didn’t miss Marisquerías la Chalana, which has a location in the vibrant and lively Plaza España.
To set the context, Spanish sidra differs from cider (or cidre) one finds in France, UK, etc. Typically a smoky flavor, what we call “funk” or “farmhouse” here in California.
Frankly, I prefer dry ciders, and the “funk” flavor is precisely what when fermenting our beloved Gravensteins from my home region — not terribly unlike Galicia or Asturias, in terms of climate and terroir. Aromatic apple varieties such as the Gravs tend to go toward smokiness quite naturally when they ferment.
So we have a range of Spanish-like sidras being produced here in west Sonoma county. Inclinado and Graviva! from Tilted Shed Ciderworks both come to mind, along with what we produce from our orchard at the Tiny Farm.
Generally speaking, Spanish sidra is a still cider, i.e., not bubbly. Spend mere moments looking at photos from sidra celebrations in Asturias and you’ll see people pouring from about a meter high, measuring from tipped bottle to tilted glass, and only serving a half-shot or so at a time. Pouring from that height aerates the still cider, which changes its flavor profile dramatically. There’s a fizz, a brighter flavor, temporarily. It’s magic — a kind of magic that dates back to ancient orchards planted by the Romans, nurtured in Spain.
So when we arrived at our table at La Chalana, Sandra and Lisa were eager to show me the robots. The restaurant group has built sidra-pouring robots — one per table, where patrons place their glass, press a button, then drink up! Serving the house sidra from Trabanco, plumbed to each table, one half-shot at a time. Drink quickly, because the magical flavor profile is temporary. And it goes so well with Spanish food!
Marco and others who grew up pouring sidra expertly since a tender young age aren’t quite convinced about the robots in this plaza (pronounced /ˈplɑːT͟Hə/) of Ciderville. Meanwhile, I’m working on my pouring technique by hand, so that we can begin to do similarly here in California as they do in Spain. Or, I’m going to drop by the local Maker space and start building sidra robots.
Other friends from Everis/NTT Data took me out to Aderezo, for an amazing fusion of regional cuisines from across Spain. Also got an excellent recommendation to check out Street XO, which is high on the list for next time I get anywhere near Madrid.
Flying back from Europe, stopping briefly at the Tiny Farm, then traveling on to Southeast Asia — though I’m not so sure anymore which of those was “West” and which was “East”? From our vantage point in Northern California, someone seems to have switched directions.
This was our third time holding Strata Data Conference in Singapore, and the quality of talks was astounding. Kudos to Ben Lorica and company who put together an excellent program. Among my favorites…
From smart cities to intelligent societies by Carmé Artigas from Synergic Partners in Madrid. In fact, it’s best to include all three of her excellent talks, especially the closing for the “Smart Cities” tutorial day. She extended beyond typical notions of smart cities, IoT, etc., into an encompassing perspective about intelligent infrastructure, and how several current emerging themes are weaving together. Her mastery of the subject most definitely changed my thinking. To paraphrase Alistair Croll’s summary at the close of the tutorial day, “Folks, that was a Master Class!” was how he described the scope of what Carmé presented. More about that in one of my upcoming talks, which has a working title of “Circularities: Evolving practices for software in the context of machine learning”.
Our very first keynote was Computational challenges and opportunities of astronomical big data by Melanie Johnston-Hollitt. If you think you work at scale, listen to what Melanie’s been building. Bigger than CERN!
Another outstanding keynote was Siri: The journey to consolidation by Cesar Delgado — which stood out to me as one of the rare moments when Apple Computer has described in detail about the scope of their use of Apache Mesos for critical infrastructure.
Also got to do some video interviews during Strata SG, and here are two of them: Adam Gibson from Skymind about using variational autoencoders for active learning; and Holden Karau from Google (plus Boo, plus another travel buddy) about PySpark and the future of Python.
Later one evening, I tagged along with Melanie, Holden, Cesar, the mellifluous and sibylline ((( matt hunt ))) (who’d just piloted some fantastic drone video of SG, legally sanctioned), Manish Bhatt (my onboarding mentor for lean start-up/continuous deployment practices at IMVU), Kirk Walter (videography maestro at O’Reilly), and a few other friends going out to The Cider Pit. Yes, in fact, there is a cider tap in Singapore. More than one. The Cider Pit serves mostly ciders and perries from the UK, and features a delicious range from Brothers. Quite surprising! The Standish retails a similar selection.
My talks were the AI within O’Reilly Media session at Strata, and also PyTextRank: Graph algorithms for enhanced natural language processing presented at the Data Science SG meetup. Many thanks for excellent questions and discussions following these. I’m super-impressed by the ML community in Singapore.
Between the session talks and appointments in the video studio, I snuck out to lunch at Din Tai Fung, a popular Taiwanese dim sum that recently moved into San Jose. Go, do not wait. Two words: chili crab. It’s a large restaurant; when I got back to the studio, two other people from our production crew had done the same :)
The way back from Singapore is generally an adventure. Literally halfway around the world for those of us who live in Northern California. Somehow it’s always quicker going out, longer returning. However, my family has been slowly migrating from California to Hawai’i — so I took time off to visit, staying with my cousin Jeff. We’ve been getting into trouble as a team since we were toddlers in a crib together. Hiking, sea kayaking, cooking, cider tasting, crashing Xmas parties in Honolulu, and generally being in tow with multiple golden retrievers.
Jeff begins taking the dogs out on his sailboat when they’re puppies. They absolutely love the water. We held a “family reunion” of sorts, with Kai (the grandpa) and his offspring and their offspring. You should’ve seen a beach full of golden retrievers when people swam by — yearning to go rescue each swimmer or surfer. Also, check out this heartfelt video about Kai’s work at Straub Medical:
I get really tired of restaurant food while traveling, anxious to get back into a kitchen — especially when there’s amazing local food available, like on Oahu. For some reason, I began obsessing about Venezuelan cachapas while still in Singapore. Probably because of two new breakfast finds there: kaya (coconut jam, similar to a lemon curd) as well as calamansi marmalade. ZOMG! One of my favorite coffeehouses / near Stanford, Coupa Café, serves cachapas and arepas on weekend,s though they take eons to fix and serve there. You can go “co-work” there while eavesdropping on several VC deals getting made at the adjacent tables.
Also, the cider in Hawai’i. Had been hearing good things about Paradise Ciders in Honolulu, which has been producing cider for nearly a year now. Out of HNL, I headed to Village Bottle Shop and Tap Room for some Lei’d Back Lilikoi by Paradise. Notably, the lilikoi — a variety of passion fruit — complements apple flavor. Paradise ferments the two together, and the resulting blend is sublimely delicious. They also have a mango/apple blend, similarly fermented together, called Mango Daze. Jeff and I tracked that down at Three Peaks in Kailua.
BTW, I really love what Stephen Manser is doing with Three Peaks: they have taps for tastings , otherwise you can only buy or refill growlers. They use a counter pressure filling system based on special growler-filling equipment from Germany, plus they use anti-microbial oxygen barrier tubing; less oxidation, far better flavor. Their selection is wide, but it’s “locals only”. Stephen is working to showcase what Hawai’i produces for craft beers, ciders, kombuchas, nitro coffees, etc. He also makes that point that growler ensure that people reuse instead of recycle. Given their island location, even the recycled beverage cans and bottles must travel by ship back to the mainland. Reuse through growlers, featuring local brews, is the sanest, most ecological approach. Given that my relatives live in Kailua, I hope to be back to visit Stephen often in this island corner of Ciderville :)
Book review for WTF? What’s the Future and Why It’s Up to Us. Full disclosure: this title was authored by my manager’s manager :)
Tim’s methodology for mapping emerging tech is really what I was hoping to get from the book. Recommended, although one really must go to the effort of applying those suggestions to the text itself, to be able to glimpse the structure of that methodology. Second-order cybernetics FTW.
The sections about some of the history of O’Reilly Media vis-a-vis emerging technology are teeming with insights that perhaps only Tim could have shared with the world. Stories about Unix, Open Source, Internet, Web 2.0, DevOps, Next:Economy, Gov 2.0, etc., should be regarded carefully by anyone working in or with technology (which, now is almost everyone), since lessons out of history tend to repeat rather often. Much like how military officers study epic poems about battles in ancient Greece. In fact, I’ve got a hunch that is partly how this book was intended — as an early Internet-era approximation of The Illiad.
For the new year, we’ve recently scheduled more instances of my online course, Get Started with NLP in Python. There’s an also advanced course in the works — similarly based on spaCy, PyTextRank, NetworkX, etc., but more along the lines of “Introduction to working with Knowledge Graphs”, and this time with group exercises. Possibly à la this research paper: Flavor network and the principles of food pairing. Or something.
I wish you all the best in 2018, and will leave you with…