And thanks to the newly released "Rogue One: A Star Wars Story", public sentiment towards droids is off the charts. It's unlikely the CNET team will stumble across anything as endearing new Star Wars droid K-2SO as it scours the halls in Las Vegas. Robots have primarily been used as marketing gimmicks or demonstration props at previous shows. That, says IHS analyst Dinesh Kithany, is set to change. "What we will see is more from the application point of view," Kithany said.

The following is the text of a talk I gave in San Francisco on December 1st, 2016. The audience was readers of my newsletter, Exponential View. You can sign up here. This is a long (7,500 word) transcript of the talk. You can scan it to see the slides and accompanying exhibits if that is easier. Or even read it in more than one sitting…. Exponential View has a purpose. In between all the emojis and all the spelling mistakes, this is what it's about: This is me on my first day at school back when I was in Zambia in sub-Saharan Africa. On the right is my friend Rehan, who I reconnected recently through Facebook. He is now known as Dr. Freeze and he does non-invasive body sculpting in Orange County. So I can get you a good rate. But I think it's important, this starting point is important. We often are inspired from where we come from and what the hell was I doing in Zambia? My dad was trained as economist and accountant, well he is retired now, but then he was an economist and was down in Zambia building the kind of institutions that we take for granted in countries like the U.S. and the U.K. to make the country function. Zambia had just got independence from the U.K. It needed a deeper civil service, it was having to build its legal system, create its system of distribution and so on. So I got an early exposure to the importance of economic institutions for making societies wealthier and making them work. While I was down in Zambia, which is a land-locked country and doesn't have great access to the sea and this is the 1970s, so we didn't have a vast range of toys.

Life has been busy for the stars of "Sherlock" since the series premiered in 2010, with Steven Moffat and Mark Gatiss applying new London style and contemporary quirks to Arthur Conan Doyle's famous consulting detective. Its fourth season -- there have been breaks -- begins Sunday on PBS' "Masterpiece: Mystery!" Martin Freeman, the series' Dr. John Watson, has gone from a guy you might have seen on the British version of "The Office" or in "The Hitchhiker's Guide to the Galaxy" to playing Bilbo Baggins in three "Hobbit" movies and the hapless Lester Nygaard in the first season of FX's "Fargo," and hosting "Saturday Night Live." Benedict Cumberbatch, its Sherlock (also in the "Hobbit" movies, as the voice of Smaug) has, among other things, played Khan in "Star Trek Into Darkness," the title role in "Doctor Strange," codebreaker Alan Turing in "The Imitation Game" and Richard III in BBC's "The Hollow Crown" Shakespeare cycle; sung "Comfortably Numb" with Pink Floyd's David Gilmour at the Royal Albert Hall; and has become something of an international, official hot guy. Conan Doyle wrote 60 Holmes stories, but the world has deemed that insufficient, and many other hands have filled out the tale. Holmes is a useful mix of specific qualities and scant details -- an attitude, occupation and method as much as a full-fleshed, full-fledged character, and so familiar that even some characters not called Sherlock Holmes, like Hugh Laurie's Dr. House on "House," are recognizably him.

Hybrid methods that utilize both content and rating information are commonly used in many recommender systems. However, most of them use either handcrafted features or the bag-of-words representation as a surrogate for the content information but they are neither effective nor natural enough. To address this problem, we develop a collaborative recurrent autoencoder (CRAE) which is a denoising recurrent autoencoder (DRAE) that models the generation of content sequences in the collaborative filtering (CF) setting. The model generalizes recent advances in recurrent deep learning from i.i.d. input to non-i.i.d. (CF-based) input and provides a new denoising scheme along with a novel learnable pooling scheme for the recurrent autoencoder. To do this, we first develop a hierarchical Bayesian model for the DRAE and then generalize it to the CF setting. The synergy between denoising and CF enables CRAE to make accurate recommendations while learning to fill in the blanks in sequences. Experiments on real-world datasets from different domains (CiteULike and Netflix) show that, by jointly modeling the order-aware generation of sequences for the content information and performing CF for the ratings, CRAE is able to significantly outperform the state of the art on both the recommendation task based on ratings and the sequence generation task based on content information.

Matching users to the right items at the right time is a fundamental task in recommendation systems. As users interact with different items over time, users' and items' feature may evolve and co-evolve over time. Traditional models based on static latent features or discretizing time into epochs can become ineffective for capturing the fine-grained temporal dynamics in the user-item interactions. We propose a coevolutionary latent feature process model that accurately captures the coevolving nature of users' and items' feature. To learn parameters, we design an efficient convex optimization algorithm with a novel low rank space sharing constraints. Extensive experiments on diverse real-world datasets demonstrate significant improvements in user behavior prediction compared to state-of-the-arts.

We learn rich natural sound representations by capitalizing on large amounts of unlabeled sound data collected in the wild. We leverage the natural synchronization between vision and sound to learn an acoustic representation using two-million unlabeled videos. Unlabeled video has the advantage that it can be economically acquired at massive scales, yet contains useful signals about natural sound. We propose a student-teacher training procedure which transfers discriminative visual knowledge from well established visual recognition models into the sound modality using unlabeled video as a bridge. Our sound representation yields significant performance improvements over the state-of-the-art results on standard benchmarks for acoustic scene/object classification. Visualizations suggest some high-level semantics automatically emerge in the sound network, even though it is trained without ground truth labels.

Collaborative filtering is a popular technique to infer users' preferences on new content based on the collective information of all users preferences. Recommender systems then use this information to make personalized suggestions to users. When users accept these recommendations it creates a feedback loop in the recommender system, and these loops iteratively influence the collaborative filtering algorithm's predictions over time. We investigate whether it is possible to identify items affected by these feedback loops. We state sufficient assumptions to deconvolve the feedback loops while keeping the inverse solution tractable. We furthermore develop a metric to unravel the recommender system's influence on the entire user-item rating matrix. We use this metric on synthetic and real-world datasets to (1) identify the extent to which the recommender system affects the final rating matrix, (2) rank frequently recommended items, and (3) distinguish whether a user's rated item was recommended or an intrinsic preference. Our results indicate that it is possible to recover the ratings matrix of intrinsic user preferences using a single snapshot of the ratings matrix without any temporal information.

We introduce the framework of {\em blind regression} motivated by {\em matrix completion} for recommendation systems: given $m$ users, $n$ movies, and a subset of user-movie ratings, the goal is to predict the unobserved user-movie ratings given the data, i.e., to complete the partially observed matrix. Following the framework of non-parametric statistics, we posit that user $u$ and movie $i$ have features $x_1(u)$ and $x_2(i)$ respectively, and their corresponding rating $y(u,i)$ is a noisy measurement of $f(x_1(u), x_2(i))$ for some unknown function $f$. In contrast with classical regression, the features $x = (x_1(u), x_2(i))$ are not observed, making it challenging to apply standard regression methods to predict the unobserved ratings. Inspired by the classical Taylor's expansion for differentiable functions, we provide a prediction algorithm that is consistent for all Lipschitz functions. In fact, the analysis through our framework naturally leads to a variant of collaborative filtering, shedding insight into the widespread success of collaborative filtering in practice. Assuming each entry is sampled independently with probability at least $\max(m^{-1+\delta},n^{-1/2+\delta})$ with $\delta > 0$, we prove that the expected fraction of our estimates with error greater than $\epsilon$ is less than $\gamma^2 / \epsilon^2$ plus a polynomially decaying term, where $\gamma^2$ is the variance of the additive entry-wise noise term. Experiments with the MovieLens and Netflix datasets suggest that our algorithm provides principled improvements over basic collaborative filtering and is competitive with matrix factorization methods.

Word embeddings are a powerful approach to capturing semantic similarity among terms in a vocabulary. In this paper, we develop exponential family embeddings, which extends the idea of word embeddings to other types of high-dimensional data. As examples, we studied several types of data: neural data with real-valued observations, count data from a market basket analysis, and ratings data from a movie recommendation system. The main idea is that each observation is modeled conditioned on a set of latent embeddings and other observations, called the context, where the way the context is defined depends on the problem. In language the context is the surrounding words; in neuroscience the context is close-by neurons; in market basket data the context is other items in the shopping cart. Each instance of an embedding defines the context, the exponential family of conditional distributions, and how the embedding vectors are shared across data. We infer the embeddings with stochastic gradient descent, with an algorithm that connects closely to generalized linear models. On all three of our applications—neural activity of zebrafish, users’ shopping behavior, and movie ratings—we found that exponential family embedding models are more effective than other dimension reduction methods. They better reconstruct held-out data and find interesting qualitative structure.

The facility location problem is widely used for summarizing large datasets and has additional applications in sensor placement, image retrieval, and clustering. One difficulty of this problem is that submodular optimization algorithms require the calculation of pairwise benefits for all items in the dataset. This is infeasible for large problems, so recent work proposed to only calculate nearest neighbor benefits. One limitation is that several strong assumptions were invoked to obtain provable approximation guarantees. In this paper we establish that these extra assumptions are not necessary--solving the sparsified problem will be almost optimal under the standard assumptions of the problem. We then analyze a different method of sparsification that is a better model for methods such as Locality Sensitive Hashing to accelerate the nearest neighbor computations and extend the use of the problem to a broader family of similarities. We validate our approach by demonstrating that it rapidly generates interpretable summaries.