Patrick Winston, a beloved professor and computer scientist at MIT, died on July 19 at Massachusetts General Hospital in Boston. A professor at MIT for almost 50 years, Winston was director of MIT's Artificial Intelligence Laboratory from 1972 to 1997 before it merged with the Laboratory for Computer Science to become MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL). A devoted teacher and cherished colleague, Winston led CSAIL's Genesis Group, which focused on developing AI systems that have human-like intelligence, including the ability to tell, perceive, and comprehend stories. He believed that such work could help illuminate aspects of human intelligence that scientists don't yet understand. "My principal interest is in figuring out what's going on inside our heads, and I'm convinced that one of the defining features of human intelligence is that we can understand stories,'" said Winston, the Ford Professor of Artificial Intelligence and Computer Science, in a 2011 interview for CSAIL.

Fernando "Corby" Corbató, an MIT professor emeritus whose work in the 1960s on time-sharing systems broke important ground in democratizing the use of computers, died on Friday, July 12, at his home in Newburyport, Massachusetts. Decades before the existence of concepts like cybersecurity and the cloud, Corbató led the development of one of the world's first operating systems. His "Compatible Time-Sharing System" (CTSS) allowed multiple people to use a computer at the same time, greatly increasing the speed at which programmers could work. It's also widely credited as the first computer system to use passwords. After CTSS Corbató led a time-sharing effort called Multics, which directly inspired operating systems like Linux and laid the foundation for many aspects of modern computing.

In this episode of Robots in Depth, Per Sjöborg speaks with Federico Pecora about AI and robotics. Federico Pecora is Associate Professor in Computer Science at the Center for Applied Autonomous Sensor Systems at Örebro University, Sweden.

In this episode, Lauren Klein interviews Michal Luria, a PhD candidate in the Human-Computer Interaction Institute at Carnegie Mellon University, about research that explores the boundaries of Human-Robot Interaction. Michal draws inspiration from the Medieval Times for her project to test how historical automata can inform modern robotics. She also discusses her work with cathartic objects to support emotional release. Michal Luria is a PhD candidate in the Human-Computer Interaction Institute at Carnegie Mellon University, advised by Professors Jodi Forlizzi and John Zimmerman. Prior to her PhD, Michal studied Interactive Communication at the Interdisciplinary Center Herzliya in Israel.

Batch-normalization (BN) layers are thought to be an integrally important layer type in today's state-of-the-art deep convolutional neural networks for computer vision tasks such as classification and detection. However, BN layers introduce complexity and computational overheads that are highly undesirable for training and/or inference on low-power custom hardware implementations of real-time embedded vision systems such as UAVs, robots and Internet of Things (IoT) devices. They are also problematic when batch sizes need to be very small during training, and innovations such as residual connections introduced more recently than BN layers could potentially have lessened their impact. In this paper we aim to quantify the benefits BN layers offer in image classification networks, in comparison with alternative choices. In particular, we study networks that use shifted-ReLU layers instead of BN layers. We found, following experiments with wide residual networks applied to the ImageNet, CIFAR 10 and CIFAR 100 image classification datasets, that BN layers do not consistently offer a significant advantage. We found that the accuracy margin offered by BN layers depends on the data set, the network size, and the bit-depth of weights. We conclude that in situations where BN layers are undesirable due to speed, memory or complexity costs, that using shifted-ReLU layers instead should be considered; we found they can offer advantages in all these areas, and often do not impose a significant accuracy cost.

The derivation of physical laws is a dominant topic in scientific research. We propose a new method capable of discovering the physical laws from data to tackle four challenges in the previous methods. The four challenges are: (1) large noise in the data, (2) outliers in the data, (3) integrating the data collected from different experiments, and (4) extrapolating the solutions to the areas that have no available data. To resolve these four challenges, we try to discover the governing differential equations and develop a model-discovering method based on sparse Bayesian inference and subsampling. The subsampling technique is used for improving the accuracy of the Bayesian learning algorithm here, while it is usually employed for estimating statistics or speeding up algorithms elsewhere. The optimal subsampling size is moderate, neither too small nor too big. Another merit of our method is that it can work with limited data by the virtue of Bayesian inference. We demonstrate how to use our method to tackle the four aforementioned challenges step by step through numerical examples: (1) predator-prey model with noise, (2) shallow water equations with outliers, (3) heat diffusion with random initial and boundary conditions, and (4) fish-harvesting problem with bifurcations. Numerical results show that the robustness and accuracy of our new method is significantly better than the other model-discovering methods and traditional regression methods.

Owe to the recent advancements in Artificial Intelligence especially deep learning, many data-driven decision support systems have been implemented to facilitate medical doctors in delivering personalized care. We focus on the deep reinforcement learning (DRL) models in this paper. DRL models have demonstrated human-level or even superior performance in the tasks of computer vision and game playings, such as Go and Atari game. However, the adoption of deep reinforcement learning techniques in clinical decision optimization is still rare. We here present the first survey that summarizes reinforcement learning algorithms with Deep Neural Networks (DNN) on clinical decision support. We also discuss some case studies, where different DRL algorithms were applied to address various clinical challenges. We further compare and contrast the advantages and limitations of various DRL algorithms and present a preliminary guide on how to choose the appropriate DRL algorithm for particular clinical applications.

It is well known that the historical logs are used for evaluating and learning policies in interactive systems, e.g. recommendation, search, and online advertising. Since direct online policy learning usually harms user experiences, it is more crucial to apply off-policy learning in real-world applications instead. Though there have been some existing works, most are focusing on learning with one single historical policy. However, in practice, usually a number of parallel experiments, e.g. multiple AB tests, are performed simultaneously. To make full use of such historical data, learning policies from multiple loggers becomes necessary. Motivated by this, in this paper, we investigate off-policy learning when the training data coming from multiple historical policies. Specifically, policies, e.g. neural networks, can be learned directly from multi-logger data, with counterfactual estimators. In order to understand the generalization ability of such estimator better, we conduct generalization error analysis for the empirical risk minimization problem. We then introduce the generalization error bound as the new risk function, which can be reduced to a constrained optimization problem. Finally, we give the corresponding learning algorithm for the new constrained problem, where we can appeal to the minimax problems to control the constraints. Extensive experiments on benchmark datasets demonstrate that the proposed methods achieve better performances than the state-of-the-arts.

Spatio-temporal (ST) data for urban applications, such as taxi demand, traffic flow, regional rainfall is inherently stochastic and unpredictable. Recently, deep learning based ST prediction models are proposed to learn the ST characteristics of data. However, it is still very challenging (1) to adequately learn the complex and non-linear ST relationships; (2) to model the high variations in the ST data volumes as it is inherently dynamic, changing over time (i.e., irregular) and highly influenced by many external factors, such as adverse weather, accidents, traffic control, PoI, etc.; and (3) as there can be many complicated external factors that can affect the accuracy and it is impossible to list them explicitly. To handle the aforementioned issues, in this paper, we propose a novel deep generative adversarial network based model (named, D-GAN) for more accurate ST prediction by implicitly learning ST feature representations in an unsupervised manner. D-GAN adopts a GAN-based structure and jointly learns generation and variational inference of data. More specifically, D-GAN consists of two major parts: (1) a deep ST feature learning network to model the ST correlations and semantic variations, and underlying factors of variations and irregularity in the data through the implicit distribution modelling; (2) a fusion module to incorporate external factors for reaching a better inference. To the best our knowledge, no prior work studies ST prediction problem via deep implicit generative model and in an unsupervised manner. Extensive experiments performed on two real-world datasets show that D-GAN achieves more accurate results than traditional as well as deep learning based ST prediction methods.

Laboratory test results are an important and generally highly dimensional component of a patient's Electronic Health Record (EHR). We train embedding representations (via Word2Vec and GloVe) for LOINC codes of laboratory tests from the EHRs of about 80,000 patients at a cancer center. To include information about lab test outcomes, we also train embeddings on the concatenation of a LOINC code with a symbol indicating normality or abnormality of the result. We observe generally clinically meaningful similarities among LOINC embeddings trained over our data. For the embeddings of the concatenation of LOINCs with abnormality codes, we evaluate the predictive performance for mortality prediction tasks and the ability to preserve ordinality properties: i.e. a lab test with normal outcome should be more similar to an abnormal one than to the a very abnormal one.