With the advent of deep neural networks, some research focuses towards understanding their black-box behavior. In this paper, we propose a new type of self-interpretable models, that are, architectures designed to provide explanations along with their predictions. Our method proceeds in two stages and is trained end-to-end: first, our model builds a low-dimensional binary representation of any input where each feature denotes the presence or absence of concepts. Then, it computes a prediction only based on this binary representation through a simple linear model. This allows an easy interpretation of the model's output in terms of presence of particular concepts in the input. The originality of our approach lies in the fact that concepts are automatically discovered at training time, without the need for additional supervision. Concepts correspond to a set of patterns, built on local low-level features (e.g a part of an image, a word in a sentence), easily identifiable from the other concepts. We experimentally demonstrate the relevance of our approach using classification tasks on two types of data, text and image, by showing its predictive performance and interpretability.

This week, I interviewed Yuval Noah Harari, the author of three best-selling books about the history and future of our species, and Fei-Fei Li, one of the pioneers in the field of artificial intelligence. The event was hosted by the Stanford Center for Ethics and Society, the Stanford Institute for Human-Centered Artificial Intelligence, and the Stanford Humanities Center. A transcript of the event follows, and a video is posted below. Nicholas Thompson: Thank you, Stanford, for inviting us all here. I want this conversation to have three parts: First, lay out where we are; then talk about some of the choices we have to make now; and last, talk about some advice for all the wonderful people in the hall. Yuval, the last time we talked, you said many, many brilliant things, but one that stuck out was a line where you said, "We are not just in a technological crisis. We are in a philosophical crisis." So explain what you meant and explain how it ties to AI. Let's get going with a note of ...

The teacher at the front of the room began his lecture half an hour ago, and he shows no signs of letting up. As he talks, he occasionally turns to write an equation on the board. "When will we ever use this?" the disengaged students ask. Next door, however, student groups have identified a real-world problem affecting the community. Teams of students are researching, exploring and using artificial intelligence to arrive at a solution.

Regularly smoking cannabis can affect teenagers so severely that they end up three years behind their classmates in terms of brain development, a landmark study has found. The results of the investigation, which involved almost 4,000 secondary school children in Canada, led researchers to conclude cannabis is more toxic for youngsters' brains than alcohol. Persistent use of the drug seriously affected basic reasoning skills – while it also had a disastrous effect on self-control, they found. Meanwhile, a separate study has found hard evidence that the main psychoactive compound in cannabis, tetrahydrocannabinol (THC), causes changes in the brain that trigger schizophrenia. In the high school study, researchers at Montreal University studied pupils from the time they entered the Canadian seventh grade – aged 12 or 13 years – for four years.

A free online version of the second edition of the book based on Stat 110, Introduction to Probability by Joe Blitzstein and Jessica Hwang, is now available here. Print copies are available via CRC Press, Amazon, and elsewhere. Stat110x is also available as an free edX course, here. The edX course focuses on animations, interactive features, readings, and problem-solving, and is complementary to the Stat 110 lecture videos on YouTube, which are available here. The Stat110x animations are available within the course and here.

Graphs arise naturally in many real-world applications including social networks, recommender systems, ontologies, biology, and computational finance. Traditionally, machine learning models for graphs have been mostly designed for static graphs. However, many applications involve evolving graphs. This introduces important challenges for learning and inference since nodes, attributes, and edges change over time. In this survey, we review the recent advances in representation learning for dynamic graphs, including dynamic knowledge graphs. We describe existing models from an encoder-decoder perspective, categorize these encoders and decoders based on the techniques they employ, and analyze the approaches in each category. We also review several prominent applications and widely used datasets, and highlight directions for future research.

Perhaps the most ambitious scientific quest in human history is the creation of general artificial intelligence, which roughly means AI that is as smart or smarter than humans. The dominant approach in the machine learning community is to attempt to discover each of the pieces required for intelligence, with the implicit assumption that some future group will complete the Herculean task of figuring out how to combine all of those pieces into a complex thinking machine. I call this the ``manual AI approach.'' This paper describes another exciting path that ultimately may be more successful at producing general AI. It is based on the clear trend in machine learning that hand-designed solutions eventually are replaced by more effective, learned solutions. The idea is to create an AI-generating algorithm (AI-GA), which automatically learns how to produce general AI. Three Pillars are essential for the approach: (1) meta-learning architectures, (2) meta-learning the learning algorithms themselves, and (3) generating effective learning environments. I argue that either approach could produce general AI first, and both are scientifically worthwhile irrespective of which is the fastest path. Because both are promising, yet the ML community is currently committed to the manual approach, I argue that our community should increase its research investment in the AI-GA approach. To encourage such research, I describe promising work in each of the Three Pillars. I also discuss AI-GA-specific safety and ethical considerations. Because it it may be the fastest path to general AI and because it is inherently scientifically interesting to understand the conditions in which a simple algorithm can produce general AI (as happened on Earth where Darwinian evolution produced human intelligence), I argue that the pursuit of AI-GAs should be considered a new grand challenge of computer science research.

Was it fair that Harry was hired but not Barry? Was it fair that Pam was fired instead of Sam? How to ensure fairness when an intelligent algorithm takes these decisions instead of a human? How to ensure that the decisions were taken based on merit and not on protected attributes like race or sex? These are the questions that must be answered now that many decisions in real life can be made through machine learning. However research in fairness of algorithms has focused on the counterfactual questions "what if?" or "why?", whereas in real life most subjective questions of consequence are contrastive: "why this but not that?". We introduce concepts and mathematical tools using causal inference to address contrastive fairness in algorithmic decision-making with illustrative thought examples.

Spoken language understanding (SLU) acts as a critical component in goal-oriented dialog systems. It typically involves identifying the speakers intent and extracting semantic slots from user utterances, which are known as intent detection (ID) and slot filling (SF). SLU problem has been intensively investigated in recent years. However, these methods just constrain SF results grammatically, solve ID and SF independently, or do not fully utilize the mutual impact of the two tasks. This paper proposes a multi-head self-attention joint model with a conditional random field (CRF) layer and a prior mask. The experiments show the effectiveness of our model, as compared with state-of-the-art models. Meanwhile, online education in China has made great progress in the last few years. But there are few intelligent educational dialog applications for students to learn foreign languages. Hence, we design an intelligent dialog robot equipped with different scenario settings to help students learn communication skills.

This paper proposes a method to gain extra supervision via multi-task learning for multi-modal video question answering. Multi-modal video question answering is an important task that aims at the joint understanding of vision and language. However, establishing large scale dataset for multi-modal video question answering is expensive and the existing benchmarks are relatively small to provide sufficient supervision. To overcome this challenge, this paper proposes a multi-task learning method which is composed of three main components: (1) multi-modal video question answering network that answers the question based on the both video and subtitle feature, (2) temporal retrieval network that predicts the time in the video clip where the question was generated from and (3) modality alignment network that solves metric learning problem to find correct association of video and subtitle modalities. By simultaneously solving related auxiliary tasks with hierarchically shared intermediate layers, the extra synergistic supervisions are provided. Motivated by curriculum learning, multi task ratio scheduling is proposed to learn easier task earlier to set inductive bias at the beginning of the training. The experiments on publicly available dataset TVQA shows state-of-the-art results, and ablation studies are conducted to prove the statistical validity.