Search for: All records

The SINDy algorithm has been successfully used to identify the governing equations of dynamical systems from time series data. In this paper, we argue that this makes SINDy a potentially useful tool for causal discovery and that existing tools for causal discovery can be used to dramatically improve the performance of SINDy as tool for robust sparse modeling and system identification. We then demonstrate empirically that augmenting the SINDy algorithm with tools from causal discovery can provides engineers with a tool for learning causally robust governing equations. 

Line charts are often used to convey high level information about time series data. Unfortunately, these charts are not always described in text, and as a result are often inaccessible to users with visual impairments who rely on screen readers. In these situations, an automated system that can describe the overall trend in a chart would be desirable. This paper presents a novel approach to classifying trends in line chart images, for use in existing chart summarization tools. Previous projects have introduced approaches to automatically summarize line charts, but have thus far been unable to describe chart trends with sufficient accuracy for real-world applications. Instead of classifying an image’s trend via a convolutional neural network (CNN) system, as has been done previously, we present an architecture similar to bag-of-words (BoW) techniques for computer vision, mapping the image classification problem to an analogous natural language problem. We divided images into matrices of image patches which we then each treated as a series of “visual words” which were used to classify each image. We utilized natural language processing (NLP) word embeddings techniques to to create embeddings of visual words that allowed us to model contextual similarity between patches. We trained a linear support vector machine (SVM) model using these patch embeddings as inputs to classify the chart trend. We compared this method against a ResNet classifier pre-trained on ImageNet. Our experimental results showed that the novel approach presented in this paper outperforms existing approaches. 

Sparsity is a desirable property as our natural environment can be described by a small number of structural primitives. Strong evidence demonstrates that the brain’s representation is both explicit and sparse, which makes it metabolically efficient by reducing the cost of code transmission. In current standardized machine learning practices, end-to-end classification pipelines are much more prevalent. For the brain, there is no single classification objective function optimized by back-propagation. Instead, the brain is highly modular and learns based on local information and learning rules. In our work, we seek to show that an unsupervised, biologically inspired sparse coding algorithm can create a sparse representation that achieves a classification accuracy on par with standard supervised learning algorithms. We leverage the concept of multi-modality to show that we can link the embedding space with multiple, heterogeneous modalities. Furthermore, we demonstrate a sparse coding model which controls the latent space and creates a sparse disentangled representation, while maintaining a high classification accuracy. 

The state-of-the-art in machine learning has been achieved primarily by deep learning artificial neural networks. These networks are powerful but biologically implausible and energy intensive. In parallel, a new paradigm of neural network is being researched that can alleviate some of the computational and energy issues. These networks, spiking neural networks (SNNs), have transformative potential if the community is able to bridge the gap between deep learning and SNNs. However, SNNs are notoriously difficult to train and lack precision in their communication. In an effort to overcome these limitations and retain the benefits of the learning process in deep learning, we investigate novel ways to translate between them. We construct several network designs with varying degrees of biological plausibility. We then test our designs on an image classification task and demonstrate our designs allow for a customized tradeoff between biological plausibility, power efficiency, inference time, and accuracy. 

Homeostatic plasticity encompasses the mechanisms by which neurons stabilize their synaptic strength and excitability in response to prolonged and destabilizing changes in their network activity. Prolonged activity blockade leads to homeostatic scaling of action potential (AP) firing rate in hippocampal neurons in part by decreased activity of N-Methyl-D-Aspartate receptors and subsequent transcriptional down-regulation of potassium channel genes includingKCNQ3which encodes K_v7.3. Neuronal K_v7 channels are mostly heterotetramers of K_v7.2 and K_v7.3 subunits and are highly enriched at the axon initial segment (AIS) where their current potently inhibits repetitive and burst firing of APs. However, whether a decrease in K_v7.3 expression occurs at the AIS during homeostatic scaling of intrinsic excitability and what signaling pathway reducesKCNQ3transcript upon prolonged activity blockade remain unknown. Here, we report that prolonged activity blockade in cultured hippocampal neurons reduces the activity of extracellular signal-regulated kinase 1/2 (ERK1/2) followed by a decrease in the activation of brain-derived neurotrophic factor (BDNF) receptor, Tropomyosin receptor kinase B (TrkB). Furthermore, both prolonged activity blockade and prolonged pharmacological inhibition of ERK1/2 decreaseKCNQ3andBDNFtranscripts as well as the density of K_v7.3 and ankyrin-G at the AIS. Collectively, our findings suggest that a reduction in the ERK1/2 activity and subsequent transcriptional down-regulation may serve as a potential signaling pathway that links prolonged activity blockade to homeostatic control of BDNF-TrkB signaling and K_v7.3 density at the AIS during homeostatic scaling of AP firing rate. 

Abstract We propose a new probabilistic programming language for the design and analysis of cyber-physical systems, especially those based on machine learning. We consider several problems arising in the design process, including training a system to be robust to rare events, testing its performance under different conditions, and debugging failures. We show how a probabilistic programming language can help address these problems by specifying distributions encoding interesting types of inputs, then sampling these to generate specialized training and test data. More generally, such languages can be used to write environment models, an essential prerequisite to any formal analysis. In this paper, we focus on systems such as autonomous cars and robots, whose environment at any point in time is ascene, a configuration of physical objects and agents. We design a domain-specific language,Scenic, for describingscenariosthat are distributions over scenes and the behaviors of their agents over time.Sceniccombines concise, readable syntax for spatiotemporal relationships with the ability to declaratively impose hard and soft constraints over the scenario. We develop specialized techniques for sampling from the resulting distribution, taking advantage of the structure provided byScenic’s domain-specific syntax. Finally, we applyScenicin multiple case studies for training, testing, and debugging neural networks for perception both as standalone components and within the context of a full cyber-physical system. 

We present a multimodal deep learning framework that can generate summarization text supporting the main idea of an information graphic for presentation to a person who is blind or visually impaired. The framework utilizes the visual, textual, positional, and size characteristics extracted from the image to create the summary. Different and complimentary neural architectures are optimized for each task using crowdsourced training data. From our quantitative experiments and results, we explain the reasoning behind our framework and show the effectiveness of our models. Our qualitative results showcase text generated from our framework and show that Mechanical Turk participants favor them to other automatic and human generated summarizations. We describe the design and results of an experiment to evaluate the utility of our system for people who have visual impairments in the context of understanding Twitter Tweets containing line graphs.