Advances in in computer vision have lead to classifiers that nearly match human performance in many applications. However, while the human visual system is remarkably versatile in extracting semantic meaning out of even degraded and heavily obfuscated images, today's visual classifiers significantly lag behind in emulating the same robustness, and often yield incorrect outputs in the presence of natural and adversarial degradations.

While the ImageNet dataset has been driving computer vision research over the past decade, significant label noise and ambiguity have made top-1 accuracy an insufficient measure of further progress.

Abstract--Autonomous robots rely on geometric maps to inform a diverse set of perception and decision-making algorithms. As autonomy requires reasoning and planning on multiple scales of the environment, each algorithm may require a different map for optimal performance. Light Detection And Ranging (LiDAR) sensors generate an abundance of geometric data to satisfy these diverse requirements, but selecting informative, size-constrained maps is computationally challenging as it requires solving an NP-hard combinatorial optimization. In this work we present OptMap: a geometric map distillation algorithm which achieves real-time, application-specific map generation via multiple theoretical and algorithmic innovations. A central feature is the maximization of set functions that exhibit diminishing returns, i.e., submodularity, using polynomial-time algorithms with provably near-optimal solutions. We formulate a novel submodular reward function which quantifies informativeness, reduces input set sizes, and minimizes bias in sequentially collected datasets. Further, we propose a dynamically reordered streaming submod-ular algorithm which improves empirical solution quality and addresses input order bias via an online approximation of the value of all scans. T esting was conducted on open-source and custom datasets with an emphasis on long-duration mapping sessions, highlighting OptMap's minimal computation requirements. Open-source ROS1 and ROS2 packages are available and can be used alongside any LiDAR SLAM algorithm. ODERN autonomous systems use a modular software architecture with separate algorithms for perceiving the environment, planning collision-free paths, estimating vehicle motion, and making higher-level decisions to complete their tasks. Many of these algorithms depend on geometric information about the environment to function properly. As a result, their performance and processing time can vary greatly depending on the quality of the geometric data. For example, trajectory planners use geometric maps to plan collision-free paths, but the density of geometric data is critical for balancing real-time replanning requirements against reliable collision detection. This trade-off is best served by dense geometric maps that specifically capture the intended trajectory corridor (Figure 1a). In contrast, localization entails aligning a source and reference point cloud, a process best served by using a sparse and global reference point could to minimize computation time while maximizing alignment accuracy (Figure 1b). Distribution Statement A: Approved for public release; distribution is unlimited. Map is dense while remaining efficient as only points near the intended trajectory are returned.

Neural Information Processing Systems http://nips.cc/

Extensive literature on backdoor poison attacks has studied attacks and defenses for backdoors using "digital trigger patterns." In contrast, "physical backdoors" use physical objects as triggers, have only recently been identified, and are qualitatively different enough to resist most defenses targeting digital trigger backdoors. Research on physical backdoors is limited by access to large datasets containing real images of physical objects co-located with misclassification targets . Building these datasets is time-and labor-intensive. This work seeks to address the challenge of accessibility for research on physical backdoor attacks.

BigGAN to learn more detailed network information. The learning rate of the generator is 0.0001, and the one of the encoder, adaptor and discriminator is 0.0004 with exponential decay We also evaluate our method using fewer animal faces. Interpolation by keeping the input image fixed while interpolating between two class embeddings. The first column is the input images, while the remaining columns are the interpolated results. Further results on the Animal faces dataset.

We explore a "best-of-both" approach to modelling molecular properties by combining learned molecular descriptors from a graph neural network (GNN) with general-purpose descriptors and a mixed ensemble of machine learning (ML) models. We introduce a MetaModel framework to aggregate predictions from a diverse set of leading ML models. We present a featurisation scheme for combining task-specific GNN-derived features with conventional molecular descriptors. We demonstrate that our framework outperforms the cutting-edge ChemProp model on all regression datasets tested and 6 of 9 classification datasets. We further show that including the GNN features derived from ChemProp boosts the ensemble model's performance on several datasets where it otherwise would have underperformed. We conclude that to achieve optimal performance across a wide set of problems, it is vital to combine general-purpose descriptors with task-specific learned features and use a diverse set of ML models to make the predictions.

Instance discrimination is a self-supervised representation learning paradigm wherein individual instances within a dataset are treated as distinct classes. This is typically achieved by generating two disparate views of each instance by applying stochastic transformations, which encourages the model to learn representations that are invariant to the common underlying object across these views. While this approach facilitates the acquisition of invariant representations for dataset instances under various handcrafted transformations (e.g., random cropping, color jittering), an exclusive reliance on such data transformations for achieving invariance may inherently limit the model's generalization to unseen datasets and diverse downstream tasks. The inherent limitation stems from the fact that the finite set of transformations within the data processing pipeline is unable to encompass the full spectrum of potential data variations. In this study, we provide the technical foundation for leveraging semantic pairs to enhance the generalization of the model's representation and empirically demonstrate that incorporating semantic pairs mitigates the issue of limited transformation coverage. Specifically, we propose that exposing the model to semantic pairs (i.e., two instances belonging to the same semantic category) introduces varied real-world scene contexts, thereby fostering the development of more generalizable object representations. To validate this hypothesis, we constructed and released a novel dataset comprising curated semantic pairs and conducted extensive experimentation to empirically establish that their inclusion enables the model to learn more general representations, ultimately leading to improved performance across diverse downstream tasks.