Network pruning has become the de facto tool to accelerate deep neural networks for mobile and edge applications. Recently, feature-map discriminant based channel pruning has shown promising results, as it aligns well with the CNN objective of differentiating multiple classes and offers better interpretability of the pruning decision. However, existing discriminant-based methods are challenged by computation inefficiency, as there is a lack of theoretical guidance on quantifying the feature-map discriminant power. In this paper, we present a new mathematical formulation to accurately and efficiently quantify the feature-map discriminativeness, which gives rise to a novel criterion,Discriminant Information(DI). We analyze the theoretical property of DI, specifically the non-decreasing property, that makes DI a valid selection criterion. DI-based pruning removes channels with minimum influence to DI value, as they contain little information regarding to the discriminant power. The versatility of DI criterion also enables an intra-layer mixed precision quantization to further compress the network. Moreover, we propose a DI-based greedy pruning algorithm and structure distillation technique to automatically decide the pruned structure that satisfies certain resource budget, which is a common requirement in reality. Extensive experiments demonstratethe effectiveness of our method: our pruned ResNet50 on ImageNet achieves 44% FLOPs reduction without any Top-1 accuracy loss compared to unpruned model

We report a triangular neural network implementation of neural autoregressive flow (NAF). Unlike many universal autoregressive density models, our design is highly modular, parameter economy, computationally efficient, and applicable to density estimation of data with high dimensions. It achieves state-of-the-art bits-per-dimension indices on MNIST and CIFAR-10 (about 1.1 and 3.7, respectively) in the category of general-purpose density estimators.

Many risk-sensitive applications require Machine Learning (ML) models to be interpretable. Attempts to obtain interpretable models typically rely on tuning, by trial-and-error, hyper-parameters of model complexity that are only loosely related to interpretability. We show that it is instead possible to take a meta-learning approach: an ML model of non-trivial Proxies of Human Interpretability (PHIs) can be learned from human feedback, then this model can be incorporated within an ML training process to directly optimize for interpretability. We show this for evolutionary symbolic regression. We first design and distribute a survey finalized at finding a link between features of mathematical formulas and two established PHIs, simulatability and decomposability. Next, we use the resulting dataset to learn an ML model of interpretability. Lastly, we query this model to estimate the interpretability of evolving solutions within bi-objective genetic programming. We perform experiments on five synthetic and eight real-world symbolic regression problems, comparing to the traditional use of solution size minimization. The results show that the use of our model leads to formulas that are, for a same level of accuracy-interpretability trade-off, either significantly more or equally accurate. Moreover, the formulas are also arguably more interpretable. Given the very positive results, we believe that our approach represents an important stepping stone for the design of next-generation interpretable (evolutionary) ML algorithms.

Convolutional Neural Networks (CNNs) have shown to be powerful medical image segmentation models. In this study, we address some of the main unresolved issues regarding these models. Specifically, training of these models on small medical image datasets is still challenging, with many studies promoting techniques such as transfer learning. Moreover, these models are infamous for producing over-confident predictions and for failing silently when presented with out-of-distribution (OOD) data at test time. In this paper, we advocate for multi-task learning, i.e., training a single model on several different datasets, spanning several different organs of interest and different imaging modalities. We show that not only a single CNN learns to automatically recognize the context and accurately segment the organ of interest in each context, but also that such a joint model often has more accurate and better-calibrated predictions than dedicated models trained separately on each dataset. Our experiments show that multi-task learning can outperform transfer learning in medical image segmentation tasks. For detecting OOD data, we propose a method based on spectral analysis of CNN feature maps. We show that different datasets, representing different imaging modalities and/or different organs of interest, have distinct spectral signatures, which can be used to identify whether or not a test image is similar to the images used to train a model. We show that this approach is far more accurate than OOD detection based on prediction uncertainty. The methods proposed in this paper contribute significantly to improving the accuracy and reliability of CNN-based medical image segmentation models.

Routing of multiple vehicles is an important and difficult problem with applications in the logistic domain [1], especially in the area of customer servicing [2]. In postal services, after-sales services, and in business to business delivery or pick up services one or more vehicles have to be efficiently routed towards customers. If customers can request services over time, the problem becomes dynamic: besides a set of fixed customers, new requests can appear at any point in time. Of course, it is desirable that as many customers as possible are serviced while the tour of any vehicle is kept short. However, it is usually infeasible (due to human resources, labor regulations, or other constraints) to service all customer requests. And clearly, the less customers are left unserviced, the longer the tours become. Thus, the problem is inherently multi-objective. Any efficient solution (smallest maximum tour across all vehicles) is a compromise between the desire to service as many customers as possible (e.g.

Deep Reinforcement Learning has been successfully applied in various computer games [8]. However, it is still rarely used in real-world applications, especially for the navigation and continuous control of real mobile robots [13]. Previous approaches lack safety and robustness and/or need a structured environment. In this paper we present our proof of concept for autonomous self-learning robot navigation in an unknown environment for a real robot without a map or planner. The input for the robot is only the fused data from a 2D laser scanner and a RGB-D camera as well as the orientation to the goal. The map of the environment is unknown. The output actions of an Asynchronous Advantage Actor-Critic network (GA3C) are the linear and angular velocities for the robot. The navigator/controller network is pretrained in a high-speed, parallel, and self-implemented simulation environment to speed up the learning process and then deployed to the real robot. To avoid overfitting, we train relatively small networks, and we add random Gaussian noise to the input laser data. The sensor data fusion with the RGB-D camera allows the robot to navigate in real environments with real 3D obstacle avoidance and without the need to fit the environment to the sensory capabilities of the robot. To further increase the robustness, we train on environments of varying difficulties and run 32 training instances simultaneously. Video: supplementary File / YouTube, Code: GitHub

There is a growing recognition that artists use valuable ways to understand and work with cognitive and perceptual mechanisms to convey desired experiences and narrative in their created artworks (DiPaola et al., 2010; Zeki, 2001). This paper documents our attempt to computationally model the creative process of a portrait painter, who relies on understanding human traits (i.e., personality and emotions) to inform their art. Our system includes an empathic conversational interaction component to capture the dominant personality category of the user and a generative AI Portraiture system that uses this categorization to create a personalized stylization of the user's portrait. This paper includes the description of our systems and the real-time interaction results obtained during the demonstration session of the NeurIPS 2019 Conference.

Tue 19 May 2020 06.14 EDT Last modified on Tue 19 May 2020 06.16 EDT When legendary chess grandmaster Garry Kasparov found himself beaten by IBM's Deep Blue supercomputer, it was seen as a seminal moment in the evolution of artificial intelligence. It was New York, 1997 and for the first time ever a computer had beaten a world champion under tournament conditions. This was the culmination of a journey in which the first stirrings of what we now call artificial intelligence and machine learning were born. A road trodden by war heroes and student researchers alike, whose singular desire to create a program that could beat the very best in the world would shape an entire science. Early origins Chess lends itself well to computer programming.

"We didn't learn the lesson on SARS and MERS," UK prime minister Boris Johnson said today as he faced questions from the House of Commons Liaison Committee, referencing the government's pandemic planning and a lack of capacity at Public Health England to detect outbreaks of coronavirus around the country. He also said that there would not be an official inquiry to investigate whether his senior aide Dominic Cummings broke lockdown rules. More than 40 Conservative party MPs have now called for Cummings' resignation. During the meeting, Johnson announced that England's test and trace system will be launched tomorrow. Under the new system, contact tracers will ask people who test positive for coronavirus to self-isolate for 14 days, regardless of symptoms, and to provide details of any recent close contacts. The secretary of state will have the power to "mandate" people to isolate if they do not isolate voluntarily. The government announced earlier today that localised lockdowns, ...

Drone firm Zipline has been given the go-ahead to deliver medical supplies and personal protective equipment to hospitals in North Carolina. The firm will be allowed to use drones on two specified routes after the Federal Aviation Administration granted it an emergency waiver. It is the first time the FAA has allowed beyond-line-of-sight drone deliveries in the US. Experts say the pandemic could help ease some drone-flight regulations. Zipline, which has been negotiating with the FAA, wants to expand to other hospitals and eventually offer deliveries to people's homes.