We study the problems of offline and online contextual optimization with feedback information, where instead of observing the loss, we observe, after-the-fact, the optimal action an oracle with full knowledge of the objective function would have taken. We aim to minimize regret, which is defined as the difference between our losses and the ones incurred by an all-knowing oracle. In the offline setting, the decision-maker has information available from past periods and needs to make one decision, while in the online setting, the decision-maker optimizes decisions dynamically over time based a new set of feasible actions and contextual functions in each period. For the offline setting, we characterize the optimal minimax policy, establishing the performance that can be achieved as a function of the underlying geometry of the information induced by the data. In the online setting, we leverage this geometric characterization to optimize the cumulative regret. We develop an algorithm that yields the first regret bound for this problem that is logarithmic in the time horizon.
A longstanding debate in semiotics centers on the relationship between linguistic signs and their corresponding semantics: is there an arbitrary relationship between a word form and its meaning, or does some systematic phenomenon pervade? For instance, does the character bigram \textit{gl} have any systematic relationship to the meaning of words like \textit{glisten}, \textit{gleam} and \textit{glow}? In this work, we offer a holistic quantification of the systematicity of the sign using mutual information and recurrent neural networks. We employ these in a data-driven and massively multilingual approach to the question, examining 106 languages. We find a statistically significant reduction in entropy when modeling a word form conditioned on its semantic representation. Encouragingly, we also recover well-attested English examples of systematic affixes. We conclude with the meta-point: Our approximate effect size (measured in bits) is quite small---despite some amount of systematicity between form and meaning, an arbitrary relationship and its resulting benefits dominate human language.
The musicological analysis of Carnatic music is challenging, owing to its rich structure and complexity. Automated \textit{r\=aga} classification, pitch detection, tonal analysis, modelling and information retrieval of this form of southern Indian classical music have, however, made significant progress in recent times. A causal analysis to investigate the musicological structure of Carnatic compositions and the identification of the relationships embedded in them have never been previously attempted. In this study, we propose a novel framework for causal discovery, using a compression-complexity measure. Owing to the limited number of compositions available, however, we generated surrogates to further facilitate the analysis of the prevailing causal relationships. Our analysis indicates that the context-free grammar, inferred from more complex compositions, such as the \textit{M\=e\d{l}akarta} \textit{r\=aga}, are a \textit{structural cause} for the \textit{Janya} \textit{r\=aga}. We also analyse certain special cases of the \textit{Janya r\=aga} in order to understand their origins and structure better.
Ensuring safe behavior for automated vehicles in unregulated traffic areas poses a complex challenge for the industry. It is an open problem to provide scalable and certifiable solutions to this challenge. We derive a trajectory planner based on model predictive control which interoperates with a monitoring system for pedestrian safety based on cellular automata. The combined planner-monitor system is demonstrated on the example of a narrow indoor parking environment. The system features deterministic behavior, mitigating the immanent risk of black boxes and offering full certifiability. By using fundamental and conservative prediction models of pedestrians the monitor is able to determine a safe drivable area in the partially occluded and unstructured parking environment. The information is fed to the trajectory planner which ensures the vehicle remains in the safe drivable area at any time through constrained optimization. We show how the approach enables solving plenty of situations in tight parking garage scenarios. Even though conservative prediction models are applied, evaluations indicate a performant system for the tested low-speed navigation.
The principle of maximum entropy is a broadly applicable technique for computing a distribution with the least amount of information possible while commonly constrained to match empirically estimated feature expectations. We seek to generalize this principle to scenarios where the empirical feature expectations cannot be computed because the model variables are only partially observed, which introduces a dependency on the learned model. Extending and generalizing the principle of latent maximum entropy, we introduce uncertain maximum entropy and describe an expectation-maximization based solution to approximately solve these problems. We show that our technique generalizes the principle of maximum entropy and latent maximum entropy and discuss a generally applicable regularization technique for adding error terms to feature expectation constraints in the event of limited data.
To improve how neural networks function it is crucial to understand their learning process. The information bottleneck theory of deep learning proposes that neural networks achieve good generalization by compressing their representations to disregard information that is not relevant to the task. However, empirical evidence for this theory is conflicting, as compression was only observed when networks used saturating activation functions. In contrast, networks with non-saturating activation functions achieved comparable levels of task performance but did not show compression. In this paper we developed more robust mutual information estimation techniques, that adapt to hidden activity of neural networks and produce more sensitive measurements of activations from all functions, especially unbounded functions. Using these adaptive estimation techniques, we explored compression in networks with a range of different activation functions. With two improved methods of estimation, firstly, we show that saturation of the activation function is not required for compression, and the amount of compression varies between different activation functions. We also find that there is a large amount of variation in compression between different network initializations. Secondary, we see that L2 regularization leads to significantly increased compression, while preventing overfitting. Finally, we show that only compression of the last layer is positively correlated with generalization.
Shape and pose estimation is a critical perception problem for a self-driving car to fully understand its surrounding environment. One fundamental challenge in solving this problem is the incomplete sensor signal (e.g., LiDAR scans), especially for faraway or occluded objects. In this paper, we propose a novel algorithm to address this challenge, we explicitly leverage the sensor signal captured over consecutive time: the consecutive signals can provide more information of an object, including different viewpoints and its motion. By encoding the consecutive signal via a recurrent neural network, our algorithm not only improves shape and pose estimation, but also produces a labeling tool that can benefit other tasks in autonomous driving research. Specifically, building upon our algorithm, we propose a novel pipeline to automatically annotate high-quality labels for amodal segmentation on images, which are hard and laborious to annotate manually. Our code and data will be made publicly available.
In this paper, we investigate the knowledge distillation (KD) strategy for object detection and propose an effective framework applicable to both homogeneous and heterogeneous student-teacher pairs. The conventional feature imitation paradigm introduces imitation masks to focus on informative foreground areas while excluding the background noises. However, we find that those methods fail to fully utilize the semantic information in all feature pyramid levels, which leads to inefficiency for knowledge distillation between FPN-based detectors. To this end, we propose a novel semantic-guided feature imitation technique, which automatically performs soft matching between feature pairs across all pyramid levels to provide the optimal guidance to the student. To push the envelop even further, we introduce contrastive distillation to effectively capture the information encoded in the relationship between different feature regions. Finally, we propose a generalized detection KD pipeline, which is capable of distilling both homogeneous and heterogeneous detector pairs. Our method consistently outperforms the existing detection KD techniques, and works when (1) components in the framework are used separately and in conjunction; (2) for both homogeneous and heterogenous student-teacher pairs and (3) on multiple detection benchmarks. With a powerful X101-FasterRCNN-Instaboost detector as the teacher, R50-FasterRCNN reaches 44.0% AP, R50-RetinaNet reaches 43.3% AP and R50-FCOS reaches 43.1% AP on COCO dataset.
Given a node-attributed graph, how can we efficiently represent it with few numerical features that expressively reflect its topology and attribute information? We propose A-DOGE, for Attributed DOS-based Graph Embedding, based on density of states (DOS, a.k.a. spectral density) to tackle this problem. A-DOGE is designed to fulfill a long desiderata of desirable characteristics. Most notably, it capitalizes on efficient approximation algorithms for DOS, that we extend to blend in node labels and attributes for the first time, making it fast and scalable for large attributed graphs and graph databases. Being based on the entire eigenspectrum of a graph, A-DOGE can capture structural and attribute properties at multiple ("glocal") scales. Moreover, it is unsupervised (i.e. agnostic to any specific objective) and lends itself to various interpretations, which makes it is suitable for exploratory graph mining tasks. Finally, it processes each graph independent of others, making it amenable for streaming settings as well as parallelization. Through extensive experiments, we show the efficacy and efficiency of A-DOGE on exploratory graph analysis and graph classification tasks, where it significantly outperforms unsupervised baselines and achieves competitive performance with modern supervised GNNs, while achieving the best trade-off between accuracy and runtime.
Modern object detectors are vulnerable to adversarial examples, which brings potential risks to numerous applications, e.g., self-driving car. Among attacks regularized by $\ell_p$ norm, $\ell_0$-attack aims to modify as few pixels as possible. Nevertheless, the problem is nontrivial since it generally requires to optimize the shape along with the texture simultaneously, which is an NP-hard problem. To address this issue, we propose a novel method of Adversarial Semantic Contour (ASC) guided by object contour as prior. With this prior, we reduce the searching space to accelerate the $\ell_0$ optimization, and also introduce more semantic information which should affect the detectors more. Based on the contour, we optimize the selection of modified pixels via sampling and their colors with gradient descent alternately. Extensive experiments demonstrate that our proposed ASC outperforms the most commonly manually designed patterns (e.g., square patches and grids) on task of disappearing. By modifying no more than 5\% and 3.5\% of the object area respectively, our proposed ASC can successfully mislead the mainstream object detectors including the SSD512, Yolov4, Mask RCNN, Faster RCNN, etc.