The seminar is a preparatory event for ICAIL 2021, with introductory talks on AI & Law. Four of the leading experts on the field have confirmed presence.

Ruth the Cookie Coach is a digital human being introduced by the Toll House brand of Nestle Global to provide baking assistance on a 24-7 basis, using an avatar incorporating AI that exhibits a degree of emotional intelligence, according to the company. Ruth is named after the creator of the Nestle Toll House original chocolate chip cookie, Ruth Wakefield. The avatar is the culmination of two years of effort between Soul Machines, which offers a Human OS platform with a Digital Brain, and Nestle. Founded in 2016 in Auckland, New Zealand, Soul Machines has raised $65 million to date, according to Crunchbase. The company was spun out of the University of Auckland by Mark Sagar, CEO and Greg Cross, chief business officer.

We report on the potential for using algorithms for non-negative matrix factorization (NMF) to improve parameter estimation in topic models. While several papers have studied connections between NMF and topic models, none have suggested leveraging these connections to develop new algorithms for fitting topic models. Importantly, NMF avoids the "sum-to-one" constraints on the topic model parameters, resulting in an optimization problem with simpler structure and more efficient computations. Building on recent advances in optimization algorithms for NMF, we show that first solving the NMF problem then recovering the topic model fit can produce remarkably better fits, and in less time, than standard algorithms for topic models. While we focus primarily on maximum likelihood estimation, we show that this approach also has the potential to improve variational inference for topic models. Our methods are implemented in the R package fastTopics.

Arguably, the visual perception of conversational agents to the physical world is a key way for them to exhibit the human-like intelligence. Image-grounded conversation is thus proposed to address this challenge. Existing works focus on exploring the multimodal dialog models that ground the conversation on a given image. In this paper, we take a step further to study image-grounded conversation under a fully open-ended setting where no paired dialog and image are assumed available. Specifically, we present Maria, a neural conversation agent powered by the visual world experiences which are retrieved from a large-scale image index. Maria consists of three flexible components, i.e., text-to-image retriever, visual concept detector and visual-knowledge-grounded response generator. The retriever aims to retrieve a correlated image to the dialog from an image index, while the visual concept detector extracts rich visual knowledge from the image. Then, the response generator is grounded on the extracted visual knowledge and dialog context to generate the target response. Extensive experiments demonstrate Maria outperforms previous state-of-the-art methods on automatic metrics and human evaluation, and can generate informative responses that have some visual commonsense of the physical world.

With the increasing popularity of graph-based methods for dimensionality reduction and representation learning, node embedding functions have become important objects of study in the literature. In this paper, we take an axiomatic approach to understanding node embedding methods, first stating three properties for embedding dissimilarity networks, then proving that all three cannot be satisfied simultaneously by any node embedding method. Similar to existing results on the impossibility of clustering under certain axiomatic assumptions, this points to fundamental difficulties inherent to node embedding tasks. Once these difficulties are identified, we then relax these axioms to allow for certain node embedding methods to be admissible in our framework.

The Olist store is an e-commerce business headquartered in Sao Paulo, Brazil. This firm acts as a single point of contact between various small businesses and the customers who wish to buy their products. Recently, they uploaded a dataset on Kaggle that contains information about 100k orders made at multiple marketplaces between 2016 to 2018. What we purchase on e-commerce websites is affected by the reviews which we read about the product posted on that website. This firm can certainly leverage these reviews to remove those products which consistently receive negative reviews.

They're called the Sampo Generation: South Koreans in their 20s and 30s who have given up (po) three (sam) of life's conventional rites of passage -- dating, marrying and having children. They've made these choices because of economic constraints and in the process have worsened South Korea's demographic imbalances. Last year, when the country registered more deaths than births for the first time in recent history, then-Vice Finance Minister Kim Yong-beom pronounced the milestone a "death cross." "I Live Alone" is one of South Korea's most popular reality TV shows. It follows the single lives of movie actors and K-pop singers engaging in mundane activities such as feeding their pets or eating ramen in the middle of the night -- all alone.

Despite the large number of sophisticated deep neural network (DNN) verification algorithms, DNN verifier developers, users, and researchers still face several challenges. First, verifier developers must contend with the rapidly changing DNN field to support new DNN operations and property types. Second, verifier users have the burden of selecting a verifier input format to specify their problem. Due to the many input formats, this decision can greatly restrict the verifiers that a user may run. Finally, researchers face difficulties in re-using benchmarks to evaluate and compare verifiers, due to the large number of input formats required to run different verifiers. Existing benchmarks are rarely in formats supported by verifiers other than the one for which the benchmark was introduced. In this work we present DNNV, a framework for reducing the burden on DNN verifier researchers, developers, and users. DNNV standardizes input and output formats, includes a simple yet expressive DSL for specifying DNN properties, and provides powerful simplification and reduction operations to facilitate the application, development, and comparison of DNN verifiers. We show how DNNV increases the support of verifiers for existing benchmarks from 30% to 74%.

While the choice of prior is one of the most critical parts of the Bayesian inference workflow, recent Bayesian deep learning models have often fallen back on vague priors, such as standard Gaussians. In this review, we highlight the importance of prior choices for Bayesian deep learning and present an overview of different priors that have been proposed for (deep) Gaussian processes, variational autoencoders, and Bayesian neural networks. We also outline different methods of learning priors for these models from data. We hope to motivate practitioners in Bayesian deep learning to think more carefully about the prior specification for their models and to provide them with some inspiration in this regard.

ACM fellow Jack Minker passed away on April 9, 2021, at the age of 93. Minker was a leader in the development of automating logistic reasoning, including deductive databases, logic programming, and artificial intelligence, but he is perhaps best known for his efforts to promote the social responsibility of scientists and human rights. In 1972, Minker was invited to join the newly constituted Committee of Concerned Scientists. He was asked to help identify Soviet computer scientists whose human rights were under attack by their government, frequently because of their career choices or because they had requested permission to emigrate from the Soviet Union. "It was something I could not refuse to do," said Jack in 2011.