This work introduces Fractional Adaptive Linear Units (FALUs), a flexible generalization of adaptive activation functions. Leveraging principles from fractional calculus, FALUs define a diverse family of activation functions (AFs) that encompass many traditional and state-of-the-art activation functions. This family includes the Sigmoid, Gaussian, ReLU, GELU, and Swish functions, as well as a large variety of smooth interpolations between these functions. Our technique requires only a small number of additional trainable parameters, and needs no further specialized optimization or initialization procedures. For this reason, FALUs present a seamless and rich automated solution to the problem of activation function optimization. Through experiments on a variety of conventional tasks and network architectures, we demonstrate the effectiveness of FALUs when compared to traditional and state-of-the-art AFs. To facilitate practical use of this work, we plan to make our code publicly available

We propose a modular encoder-decoder model framework, NEED (NAS-Enhanced Encoder-Decoder), for interactive image segmentation. As part of our solution, we introduce a general-purpose convolutional module, Inception-Atrous-Collation (IAC) blocks. IAC structures collate multiple kernel and dilation sizes with residual information from previous network layers to generate information-rich image features. Our network additionally incorporates “context-aware” skip connections, which serve to amplify the influence of the user-provided interactive features throughout the model. With this general structure in place, we optimize our final network architecture using a Neural Architecture Search (NAS) procedure. The resultant model is a lightweight network that outperforms state-of-the-art algorithms with respect to both overall segmentation accuracy (IOU) and fine-grain boundary segmentation accuracy (bIOU) - a metric we introduce. We provide experimental results across several standard datasets, in addition to ablation studies to validate the effectiveness of our algorithm.

We present a robust, parts-based data compression algorithm, L21 Semi-Nonnegative Matrix Factorization (L21 SNF) for mixed-sign data. To resolve the instability issue caused by the Frobenius norm due to the effects of outliers, we utilize the noise-free L2,1 norm and a regularization parameter in our algorithm. We derive a rigorous proof of convergence of our algorithm. Based on experiments on large-scale over-determined matrices and real facial image data, L21 SNF demonstrates a significant improvement in accuracy over other classical methods. Furthermore, L21 SNF has a simple programming structure and can be implemented within data compression software for compression of highly over-determined systems encountered broadly across many general machine learning processes.

We apply two evolutionary search algorithms: Particle Swarm Optimization (PSO) and Genetic Algorithms (GAs) to the design of Cellular Automata (CA) that can perform computational tasks requiring global coordination. In particular, we compare search efficiency for PSO and GAs applied to both the density classification problem and to the novel generation of ”chaotic” CA. Our work furthermore introduces a new variant of PSO, the Binary Global-Local PSO (BGL-PSO).

We devise an algorithm using a Bayesian optimization framework in conjunction with contextual visual data for the efficient localization of objects in still images. Recent research has demonstrated substantial progress in object localization and related tasks for computer vision. However, many current state-of-the-art object localization procedures still suffer from inaccuracy and inefficiency, in addition to failing to provide a principled and interpretable system amenable to high-level vision tasks. We address these issues with the current research. Our method encompasses an active search procedure that uses contextual data to generate initial bounding-box proposals for a target object. We train a convolutional neural network to approximate an offset distance from the target object. Next, we use a Gaussian Process to model this offset response signal over the search space of the target. We then employ a Bayesian active search for accurate localization of the target. In experiments, we compare our approach to a state-of-the-art bounding-box regression method for a challenging pedestrian localization task. Our method exhibits a substantial improvement over this baseline regression method.

A major goal of computer vision is to enable computers to interpret visual situations - abstract concepts (e.g., “a person walking a dog,” “a crowd waiting for a bus,” “a picnic”) whose image instantiations are linked more by their common spatial and semantic structure than by low-level visual similarity. In this paper, we propose a novel method for prior learning and active object localization for this kind of knowledge-driven search in static images. In our system, prior situation knowledge is captured by a set of flexible, kernel-based density estimations - a situation model - that represent the expected spatial structure of the given situation. These estimations are updated by information gained as the system searches for relevant objects, allowing the system to use context as it is discovered to narrow the search. More specifically, at any given time in a run on a test image, our system uses image features plus contextual information it has discovered to identify a small subset of training images - an importance cluster - that is deemed most similar to the given test image, given the context. This subset is used to generate an updated situation model in an on-line fashion, using an efficient multipole expansion technique. As a proof of concept, we apply our algorithm to a highly varied and challenging dataset consisting of instances of a “dogwalking” situation. Our results support the hypothesis that dynamically-rendered, context-based probabilistic models can support efficient object localization in visual situations. Moreover, our approach is general enough to be applied to diverse machine learning paradigms requiring interpretable, probabilistic representations generated from partially observed data.