Search for: All records

We introduce a convolutional neural network model for unsupervised learning of depth and ego-motion from cylindrical panoramic video. Panoramic depth estimation is an important technology for applications such as virtual reality, 3d modeling, and autonomous robotic navigation. In contrast to previous approaches for applying convolutional neural networks to panoramic imagery, we use the cylindrical panoramic projection which allows for the use of the traditional CNN layers such as convolutional filters and max pooling without modification. Our evaluation of synthetic and real data shows that unsupervised learning of depth and ego-motion on cylindrical panoramic images can produce high-quality depth maps and that an increased field-of-view improves ego-motion estimation accuracy. We also introduce Headcam, a novel dataset of panoramic video collected from a helmet-mounted camera while biking in an urban setting. 

Given vector representations for individual words, it is necessary to compute vector representations of sentences for many applications in a compositional manner, often using artificial neural networks. Relatively little work has explored the internal structure and properties of such sentence vectors. In this paper, we explore the properties of sentence vectors in the context of automatic summarization. In particular, we show that cosine similarity between sentence vectors and document vectors is strongly correlated with sentence importance and that vector semantics can identify and correct gaps between the sentences chosen so far and the document. In addition, we identify specific dimensions which are linked to effective summaries. To our knowledge, this is the first time specific dimensions of sentence embeddings have been connected to sentence properties. We also compare the features of different methods of sentence embeddings. Many of these insights have applications in uses of sentence embeddings far beyond summarization. 

Many challenges in natural language pro- cessing require generating text, including language translation, dialogue generation, and speech recognition. For all of these problems, text generation becomes more difficult as the text becomes longer. Cur- rent language models often struggle to keep track of coherence for long pieces of text. Here, we attempt to have the model construct and use an outline of the text it generates to keep it focused. We find that the usage of an outline improves perplex- ity. We do not find that using the outline improves human evaluation over a simpler baseline, revealing a discrepancy in per- plexity and human perception. Similarly, hierarchical generation is not found to im- prove human evaluation scores. 

Neural models enjoy widespread use across a variety of tasks and have grown to become crucial components of many industrial systems. Despite their effectiveness and ex- tensive popularity, they are not without their exploitable flaws. Initially applied to computer vision systems, the generation of adversarial examples is a process in which seemingly imper- ceptible perturbations are made to an image, with the purpose of inducing a deep learning based classifier to misclassify the image. Due to recent trends in speech processing, this has become a noticeable issue in speech recognition models. In late 2017, an attack was shown to be quite effective against the Speech Commands classification model. Limited-vocabulary speech classifiers, such as the Speech Commands model, are used quite frequently in a variety of applications, particularly in managing automated attendants in telephony contexts. As such, adversarial examples produced by this attack could have real-world consequences. While previous work in defending against these adversarial examples has investigated using audio preprocessing to reduce or distort adversarial noise, this work explores the idea of flooding particular frequency bands of an audio signal with random noise in order to detect adversarial examples. This technique of flooding, which does not require retraining or modifying the model, is inspired by work done in computer vision and builds on the idea that speech classifiers are relatively robust to natural noise. A combined defense incorporating 5 different frequency bands for flooding the signal with noise outperformed other existing defenses in the audio space, detecting adversarial examples with 91.8% precision and 93.5% recall. 

Although people have the ability to en- gage in vapid dialogue without effort, this may not be a uniquely human trait. Since the 1960’s researchers have been trying to create agents that can generate artificial conversation. These programs are com- monly known as chatbots. With increasing use of neural networks for dialog genera- tion, some conclude that this goal has been achieved. This research joins the quest by creating a dialog generating Recurrent Neural Network (RNN) and by enhancing the ability of this network with auxiliary loss functions and a beam search. Our cus- tom loss functions achieve better cohesion and coherence by including calculations of Maximum Mutual Information (MMI) and entropy. We demonstrate the effectiveness of this system by using a set of custom evaluation metrics inspired by an abun- dance of previous research and based on tried-and-true principles of Natural Lan- guage Processing. 

Recent papers in neural machine translation have proposed the strict use of attention mechanisms over previous stan- dards such as recurrent and convolutional neural networks (RNNs and CNNs). We propose that by running traditionally stacked encoding branches from encoder-decoder attention- focused architectures in parallel, that even more sequential operations can be removed from the model and thereby de- crease training time. In particular, we modify the recently published attention-based architecture called Transformer by Google, by replacing sequential attention modules with par- allel ones, reducing the amount of training time and substan- tially improving BLEU scores at the same time. Experiments over the English to German and English to French translation tasks show that our model establishes a new state of the art. 

In a world of proliferating data, the abil- ity to rapidly summarize text is grow- ing in importance. Automatic summariza- tion of text can be thought of as a se- quence to sequence problem. Another area of natural language processing that solves a sequence to sequence problem is ma- chine translation, which is rapidly evolv- ing due to the development of attention- based encoder-decoder networks. This work applies these modern techniques to abstractive summarization. We perform analysis on various attention mechanisms for summarization with the goal of devel- oping an approach and architecture aimed at improving the state of the art. In par- ticular, we modify and optimize a trans- lation model with self-attention for gener- ating abstractive sentence summaries. The effectiveness of this base model along with attention variants is compared and ana- lyzed in the context of standardized eval- uation sets and test metrics. However, we show that these metrics are limited in their ability to effectively score abstractive summaries, and propose a new approach based on the intuition that an abstractive model requires an abstractive evaluation. 

An adversarial attack is an exploitative process in which minute alterations are made to natural inputs, causing the inputs to be misclassified by neural models. In the field of speech recognition, this has become an issue of increasing significance. Although adversarial attacks were originally introduced in computer vision, they have since infiltrated the realm of speech recognition. In 2017, a genetic attack was shown to be quite potent against the Speech Commands Model. Limited-vocabulary speech classifiers, such as the Speech Commands Model, are used in a variety of applications, particularly in telephony; as such, adversarial examples produced by this attack pose as a major security threat. This paper explores various methods of detecting these adversarial examples with combinations of audio preprocessing. One particular combined defense incorporating compressions, speech coding, filtering, and audio panning was shown to be quite effective against the attack on the Speech Commands Model, detecting audio adversarial examples with 93.5% precision and 91.2% recall. 

We propose the use of dilated filters to construct an aggregation module in a multicolumn convolutional neural network for perspective-free counting. Counting is a common problem in computer vision (e.g. traffic on the street or pedestrians in a crowd). Modern approaches to the counting problem involve the production of a density map via regression whose integral is equal to the number of objects in the image. However, objects in the image can occur at different scales (e.g. due to perspective effects) which can make it difficult for a learning agent to learn the proper density map. While the use of multiple columns to extract multiscale information from images has been shown before, our approach aggregates the multiscale information gathered by the multicolumn convolutional neural network to improve performance. Our experiments show that our proposed network outperforms the state-of-the-art on many benchmark datasets, and also that using our aggregation module in combination with a higher number of columns is beneficial for multiscale counting. 

Complex neural network architectures are being increasingly used to learn to compute the semantic resemblances among natural language texts. It is necessary to establish a lower bound of performance that must be met in or- der for new complex architectures to be not only novel, but also worthwhile in terms of implementation. This paper focuses on the specific task of determin- ing semantic textual similarity (STS). We construct a number of models from simple to complex within a framework and report our results. Our findings show that a small number of LSTM stacks with an LSTM stack comparator produces the best results. We use Se- mEval 2017 STS Competition Dataset for evaluation.