We introduce a new notion of generalization -- Distributional Generalization -- which roughly states that outputs of a classifier at train and test time are close *as distributions*, as opposed to close in just their average error. For example, if we mislabel 30% of dogs as cats in the train set of CIFAR-10, then a ResNet trained to interpolation will in fact mislabel roughly 30% of dogs as cats on the *test set* as well, while leaving other classes unaffected. This behavior is not captured by classical generalization, which would only consider the average error and not the distribution of errors over the input domain. Our formal conjectures, which are much more general than this example, characterize the form of distributional generalization that can be expected in terms of problem parameters: model architecture, training procedure, number of samples, and data distribution. We give empirical evidence for these conjectures across a variety of domains in machine learning, including neural networks, kernel machines, and decision trees. Our results thus advance our empirical understanding of interpolating classifiers.

Social insect colonies routinely face large vertebrate predators, against which they need to mount a collective defense. To do so, honeybees use an alarm pheromone that recruits nearby bees into mass stinging of the perceived threat. This alarm pheromone is carried directly on the stinger, hence its concentration builds up during the course of the attack. Here, we investigate how individual bees react to different alarm pheromone concentrations, and how this evolved response-pattern leads to better coordination at the group level. We first present an individual dose-response curve to the alarm pheromone, obtained experimentally. Second, we apply Projective Simulation to model each bee as an artificial learning agent that relies on the pheromone concentration to decide whether to sting or not. If the emergent collective performance benefits the colony, the individual reactions that led to it are enhanced via reinforcement learning, thus emulating natural selection. Predators are modeled in a realistic way so that the effect of factors such as their resistance, their killing rate or their frequency of attacks can be studied. We are able to reproduce the experimentally measured response-pattern of real bees, and to identify the main selection pressures that shaped it. Finally, we apply the model to a case study: by tuning the parameters to represent the environmental conditions of European or African bees, we can predict the difference in aggressiveness observed between these two subspecies.

Caterpillar's autonomous driving technology, which can be bolted on to existing machines, is helping the U.S. heavy equipment maker mitigate the heavy impact of the coronavirus crisis on sales of its traditional workhorses. With both small and large customers looking to protect their operations from future disruptions, demand has surged for machines that don't require human operators on board. Sales of Caterpillar's autonomous technology for mining operations have been growing at a double-digit percentage clip this year compared with 2019, according to previously unreported internal company data shared with Reuters. Fred Rio, worldwide product manager at Caterpillar's construction digital & technology division, told Reuters that a remote-control technology, which allows users to operate machines from several miles away, would be available for construction sites in January. The company is also working with space agencies to use satellite technology to allow an operator sitting in the United States to remotely communicate with machines on job sites in, say, Africa or elsewhere in the world, he said.

CHICAGO (Reuters) - Question: How can a company like Caterpillar CAT.N try to counter a slump in sales of bulldozers and trucks during a pandemic that has made every human a potential disease vector? Caterpillar's autonomous driving technology, which can be bolted on to existing machines, is helping the U.S. heavy equipment maker mitigate the heavy impact of the coronavirus crisis on sales of its traditional workhorses. With both small and large customers looking to protect their operations from future disruptions, demand has surged for machines that don't require human operators on board. Sales of Caterpillar's autonomous technology for mining operations have been growing at a double-digit percentage clip this year compared with 2019, according to previously unreported internal company data shared with Reuters. By contrast, sales of its yellow bulldozers, mining trucks and other equipment have been falling for the past nine months, a trend that's also hit its main rivals including Japan's Komatsu Ltd 6301.T and American player Deere & Co DE.N .

The Machine Learning for Mobile Robot Navigation in the Wild Symposium will consist of invited talks, technical presentations, spotlight posters, robot demonstrations, industry spotlights, breakout sessions, and interactive panel discussions. All contributions should be submitted electronically via AAAI EasyChair site.

Every lion has its own unique roar, one that lets the'kings of the jungle' recognise each other and could be used to track population movements, a study has found. Researchers from Oxford used machine learning to analyse the roars of various lions -- picking out the distinguishing frequency that can be used to tell them apart. According to the experts, lions' calls are usually issued in a set -- with one or two soft moans followed by several loud, full-throated roars and finishing with grunts. Previous research had suggested that lions could distinguish their peer's roars from each other -- allowing them to identify distant friends and hostile neighbours. However, it had not previously been clear what aspects of the calls' structure allowed them to discriminate between those made by different individuals.

The recently proposed Activation Relaxation (AR) algorithm provides a simple and robust approach for approximating the backpropagation of error algorithm using only local learning rules. We have previously shown that the algorithm can be further simplified and made more biologically plausible by (i) introducing a learnable set of backwards weights, which overcomes the weight-transport problem, and (ii) avoiding the computation of nonlinear derivatives at each neuron. However, tthe efficacy of these simplifications has, so far, only been tested on simple multi-layer-perceptron (MLP) networks. Here, we show that these simplifications still maintain performance using more complex CNN architectures and challenging datasets, which have proven difficult for other biologically-plausible schemes to scale to. We also investigate whether another biologically implausible assumption of the original AR algorithm - the frozen feedforward pass - can be relaxed without damaging performance. The backpropagation of error algorithm (backprop) has been the engine driving the successes of modern machine learning with deep neural networks.

While a diverse collection of continual learning (CL) methods has been proposed to prevent catastrophic forgetting, a thorough investigation of their effectiveness for processing sequential data with recurrent neural networks (RNNs) is lacking. Here, we provide the first comprehensive evaluation of established CL methods on a variety of sequential data benchmarks. Specifically, we shed light on the particularities that arise when applying weight-importance methods, such as elastic weight consolidation, to RNNs. In contrast to feedforward networks, RNNs iteratively reuse a shared set of weights and require working memory to process input samples. We show that the performance of weight-importance methods is not directly affected by the length of the processed sequences, but rather by high working memory requirements, which lead to an increased need for stability at the cost of decreased plasticity for learning subsequent tasks. We additionally provide theoretical arguments supporting this interpretation by studying linear RNNs. Our study shows that established CL methods can be successfully ported to the recurrent case, and that a recent regularization approach based on hypernetworks outperforms weight-importance methods, thus emerging as a promising candidate for CL in RNNs. Overall, we provide insights on the differences between CL in feedforward networks and RNNs, while guiding towards effective solutions to tackle CL on sequential data.

Classic decision tree learning is a binary classification algorithm that constructs models with first-class transparency - every classification has a directly derivable explanation. However, learning decision trees on modern datasets generates large trees, which in turn generate decision paths of excessive depth, obscuring the explanation of classifications. To improve the comprehensibility of classifications, we propose a new decision tree model that we call Cascading Decision Trees. Cascading Decision Trees shorten the size of explanations of classifications, without sacrificing model performance overall. Our key insight is to separate the notion of a decision path and an explanation path. Utilizing this insight, instead of having one monolithic decision tree, we build several smaller decision subtrees and cascade them in sequence. Our cascading decision subtrees are designed to specifically target explanations for positive classifications. This way each subtree identifies the smallest set of features that can classify as many positive samples as possible, without misclassifying any negative samples. Applying cascading decision trees to new samples results in a significantly shorter and succinct explanation, if one of the subtrees detects a positive classification. In that case, we immediately stop and report the decision path of only the current subtree to the user as an explanation for the classification. We evaluate our algorithm on standard datasets, as well as new real-world applications and find that our model shortens the explanation depth by over 40.8% for positive classifications compared to the classic decision tree model.

Building destruction during war is a specific form of violence which is particularly harmful to civilians, commonly used to displace populations, and therefore warrants special attention. Yet, data from war-ridden areas are typically scarce, often incomplete and highly contested, when available. The lack of such data from conflict zones severely limits media reporting, humanitarian relief efforts, human rights monitoring, reconstruction initiatives, as well as the study of violent conflict in academic research. One approach has been to use remote sensing to identify destruction in satellite images[1]. This approach is gaining momentum as high-resolution imagery is becoming readily available and is updated ever quicker yielding weekly or even daily frequency. At the same time recent methodological advances related to deep learning have provided sophisticated tools to extract data from these images [2, 3, 4, 5].