Autonomous systems that are out in the real world today deal with a slew of different data modalities to perform effectively in tasks ranging from robot navigation in complex maneuverable robots to identity verification in simpler static systems. The performance of the system heavily banks on the continuous supply of data from all modalities. These systems can face drastically increased risk with the loss of one or multiple modalities due to an adverse scenario like that of hardware malfunction, inimical environmental conditions, etc. This thesis investigates modality hallucination and its efficacy in mitigating the risks posed to the autonomous system. Modality hallucination is proposed as one effective way to ensure consistent modality availability thereby reducing unfavorable consequences. While there has been a significant research effort in high-to-low dimensional modality hallucination, like that of RGB to depth, there is considerably lesser interest in the other direction( low-to-high dimensional modality prediction). This thesis serves to demonstrate the effectiveness of this low-to-high modality hallucination in reducing the uncertainty in the affected system while also ensuring that the method remains task agnostic.

A deep neural network based encoder-decoder architecture that aggregates multiple fields of view in its encoder blocks to recover the lost information of the affected modality from the extant modality is presented with evidence of its efficacy. The hallucination process is implemented by capturing a non-linear mapping between the data modalities and the learned mapping is used to aid the extant modality to mitigate the risk posed to the system in the adverse scenarios which involve modality loss. The results are compared with a well known generative model built for the task of image translation, as well as an off-the-shelf semantic segmentation architecture re-purposed for hallucination. To validate the practicality of hallucinated modality, extensive classification and segmentation experiments are conducted on the University of Washington's depth image database (UWRGBD) database and the New York University database (NYUD) and demonstrate that hallucination indeed lessens the negative effects of the modality loss.

In this thesis, I present two new datasets and a modification to the existing models in the form of a novel attention mechanism for Natural Language Inference (NLI). The new datasets have been carefully synthesized from various existing corpora released for different tasks.

The task of NLI is to determine the possibility of a sentence referred to as “Hypothesis” being true given that another sentence referred to as “Premise” is true. In other words, the task is to identify whether the “Premise” entails, contradicts or remains neutral with regards to the “Hypothesis”. NLI is a precursor to solving many Natural Language Processing (NLP) tasks such as Question Answering and Semantic Search. For example, in Question Answering systems, the question is paraphrased to form a declarative statement which is treated as the hypothesis. The options are treated as the premise. The option with the maximum entailment score is considered as the answer. Considering the applications of NLI, the importance of having a strong NLI system can't be stressed enough.

Many large-scale datasets and models have been released in order to advance the field of NLI. While all of these models do get good accuracy on the test sets of the datasets they were trained on, they fail to capture the basic understanding of “Entities” and “Roles”. They often make the mistake of inferring that “John went to the market.” from “Peter went to the market.” failing to capture the notion of “Entities”. In other cases, these models don't understand the difference in the “Roles” played by the same entities in “Premise” and “Hypothesis” sentences and end up wrongly inferring that “Peter drove John to the stadium.” from “John drove Peter to the stadium.”

The lack of understanding of “Roles” can be attributed to the lack of such examples in the various existing datasets. The reason for the existing model’s failure in capturing the notion of “Entities” is not just due to the lack of such examples in the existing NLI datasets. It can also be attributed to the strict use of vector similarity in the “word-to-word” attention mechanism being used in the existing architectures.

To overcome these issues, I present two new datasets to help make the NLI systems capture the notion of “Entities” and “Roles”. The “NER Changed” (NC) dataset and the “Role-Switched” (RS) dataset contains examples of Premise-Hypothesis pairs that require the understanding of “Entities” and “Roles” respectively in order to be able to make correct inferences. This work shows how the existing architectures perform poorly on the “NER Changed” (NC) dataset even after being trained on the new datasets. In order to help the existing architectures, understand the notion of “Entities”, this work proposes a modification to the “word-to-word” attention mechanism. Instead of relying on vector similarity alone, the modified architectures learn to incorporate the “Symbolic Similarity” as well by using the Named-Entity features of the Premise and Hypothesis sentences. The new modified architectures not only perform significantly better than the unmodified architectures on the “NER Changed” (NC) dataset but also performs as well on the existing datasets.

The ubiquity of single camera systems in society has made improving monocular depth estimation a topic of increasing interest in the broader computer vision community. Inspired by recent work in sparse-to-dense depth estimation, this thesis focuses on sparse patterns generated from feature detection based algorithms as opposed to regular grid sparse patterns used by previous work. This work focuses on using these feature-based sparse patterns to generate additional depth information by interpolating regions between clusters of samples that are in close proximity to each other. These interpolated sparse depths are used to enforce additional constraints on the network’s predictions. In addition to the improved depth prediction performance observed from incorporating the sparse sample information in the network compared to pure RGB-based methods, the experiments show that actively retraining a network on a small number of samples that deviate most from the interpolated sparse depths leads to better depth prediction overall.

This thesis also introduces a new metric, titled Edge, to quantify model performance in regions of an image that show the highest change in ground truth depth values along either the x-axis or the y-axis. Existing metrics in depth estimation like Root Mean Square Error(RMSE) and Mean Absolute Error(MAE) quantify model performance across the entire image and don’t focus on specific regions of an image that are hard to predict. To this end, the proposed Edge metric focuses specifically on these hard to classify regions. The experiments also show that using the Edge metric as a small addition to existing loss functions like L1 loss in current state-of-the-art methods leads to vastly improved performance in these hard to classify regions, while also improving performance across the board in every other metric.

Feature embeddings differ from raw features in the sense that the former obey certain properties like notion of similarity/dissimilarity in it's embedding space. word2vec is a preeminent example in this direction, where the similarity in the embedding space is measured in terms of the cosine similarity. Such language embedding models have seen numerous applications in both language and vision community as they capture the information in the modality (English language) efficiently. Inspired by these language models, this work focuses on learning embedding spaces for two visual computing tasks, 1. Image Hashing 2. Zero Shot Learning. The training set was used to learn embedding spaces over which similarity/dissimilarity is measured using several distance metrics like hamming / euclidean / cosine distances. While the above-mentioned language models learn generic word embeddings, in this work task specific embeddings were learnt which can be used for Image Retrieval and Classification separately.

Image Hashing is the task of mapping images to binary codes such that some notion of user-defined similarity is preserved. The first part of this work focuses on designing a new framework that uses the hash-tags associated with web images to learn the binary codes. Such codes can be used in several applications like Image Retrieval and Image Classification. Further, this framework requires no labelled data, leaving it very inexpensive. Results show that the proposed approach surpasses the state-of-art approaches by a significant margin.

Zero-shot classification is the task of classifying the test sample into a new class which was not seen during training. This is possible by establishing a relationship between the training and the testing classes using auxiliary information. In the second part of this thesis, a framework is designed that trains using the handcrafted attribute vectors and word vectors but doesn’t require the expensive attribute vectors during test time. More specifically, an intermediate space is learnt between the word vector space and the image feature space using the hand-crafted attribute vectors. Preliminary results on two zero-shot classification datasets show that this is a promising direction to explore.

Reasoning with commonsense knowledge is an integral component of human behavior. It is due to this capability that people know that a weak person may not be able to lift someone. It has been a long standing goal of the Artificial Intelligence community to simulate such commonsense reasoning abilities in machines. Over the years, many advances have been made and various challenges have been proposed to test their abilities. The Winograd Schema Challenge (WSC) is one such Natural Language Understanding (NLU) task which was also proposed as an alternative to the Turing Test. It is made up of textual question answering problems which require resolution of a pronoun to its correct antecedent.

In this thesis, two approaches of developing NLU systems to solve the Winograd Schema Challenge are demonstrated. To this end, a semantic parser is presented, various kinds of commonsense knowledge are identified, techniques to extract commonsense knowledge are developed and two commonsense reasoning algorithms are presented. The usefulness of the developed tools and techniques is shown by applying them to solve the challenge.

91% of smartphone and tablet users experience a problem with their device screen being oriented the wrong way during use [11]. In [11], the authors proposed iRotate, a previous solution which uses computer vision to solve the orientation problem. We propose iLieDown, an improved method of automatically rotating smartphones, tablets, and other device displays. This paper introduces a new algorithm to correctly orient the display relative to the user’s face using a convolutional neural network (CNN). The CNN model is trained to predict the rotation of faces in various environments through data augmentation, uses a confidence threshold, and analyzes multiple images to be accurate and robust. iLieDown is battery and CPU efficient, causes no noticeable lag to the user during use, and is 6x more accurate than iRotate.

Rapid growth of internet and connected devices ranging from cloud systems to internet of things have raised critical concerns for securing these systems. In the recent past, security attacks on different kinds of devices have evolved in terms of complexity and diversity. One of the challenges is establishing secure communication in the network among various devices and systems. Despite being protected with authentication and encryption, the network still needs to be protected against cyber-attacks. For this, the network traffic has to be closely monitored and should detect anomalies and intrusions. Intrusion detection can be categorized as a network traffic classification problem in machine learning. Existing network traffic classification methods require a lot of training and data preprocessing, and this problem is more serious if the dataset size is huge. In addition, the machine learning and deep learning methods that have been used so far were trained on datasets that contain obsolete attacks. In this thesis, these problems are addressed by using ensemble methods applied on an up to date network attacks dataset. Ensemble methods use multiple learning algorithms to get better classification accuracy that could be obtained when the corresponding learning algorithm is applied alone. This dataset for network traffic classification has recent attack scenarios and contains over fifteen attacks. This approach shows that ensemble methods can be used to classify network traffic and detect intrusions with less training times of the model, and lesser pre-processing without feature selection. In addition, this thesis also shows that only with less than ten percent of the total features of input dataset will lead to similar accuracy that is achieved on whole dataset. This can heavily reduce the training times and classification duration in real-time scenarios.

Rapid intraoperative diagnosis of brain tumors is of great importance for planning treatment and guiding the surgeon about the extent of resection. Currently, the standard for the preliminary intraoperative tissue analysis is frozen section biopsy that has major limitations such as tissue freezing and cutting artifacts, sampling errors, lack of immediate interaction between the pathologist and the surgeon, and time consuming.

Handheld, portable confocal laser endomicroscopy (CLE) is being explored in neurosurgery for its ability to image histopathological features of tissue at cellular resolution in real time during brain tumor surgery. Over the course of examination of the surgical tumor resection, hundreds to thousands of images may be collected. The high number of images requires significant time and storage load for subsequent reviewing, which motivated several research groups to employ deep convolutional neural networks (DCNNs) to improve its utility during surgery. DCNNs have proven to be useful in natural and medical image analysis tasks such as classification, object detection, and image segmentation.

This thesis proposes using DCNNs for analyzing CLE images of brain tumors. Particularly, it explores the practicality of DCNNs in three main tasks. First, off-the shelf DCNNs were used to classify images into diagnostic and non-diagnostic. Further experiments showed that both ensemble modeling and transfer learning improved the classifier’s accuracy in evaluating the diagnostic quality of new images at test stage. Second, a weakly-supervised learning pipeline was developed for localizing key features of diagnostic CLE images from gliomas. Third, image style transfer was used to improve the diagnostic quality of CLE images from glioma tumors by transforming the histology patterns in CLE images of fluorescein sodium-stained tissue into the ones in conventional hematoxylin and eosin-stained tissue slides.

These studies suggest that DCNNs are opted for analysis of CLE images. They may assist surgeons in sorting out the non-diagnostic images, highlighting the key regions and enhancing their appearance through pattern transformation in real time. With recent advances in deep learning such as generative adversarial networks and semi-supervised learning, new research directions need to be followed to discover more promises of DCNNs in CLE image analysis.

A defense-by-randomization framework is proposed as an effective defense mechanism against different types of adversarial attacks on neural networks. Experiments were conducted by selecting a combination of differently constructed image classification neural networks to observe which combinations applied to this framework were most effective in maximizing classification accuracy. Furthermore, the reasons why particular combinations were more effective than others is explored.

Propaganda bots are malicious bots on Twitter that spread divisive opinions and support political accounts. This project is based on detecting propaganda bots on Twitter using machine learning. Once I began to observe patterns within propaganda followers on Twitter, I determined that I could train algorithms to detect these bots. The paper focuses on my development and process of training classifiers and using them to create a user-facing server that performs prediction functions automatically. The learning goals of this project were detailed, the focus of which was to learn some form of machine learning architecture. I needed to learn some aspect of large data handling, as well as being able to maintain these datasets for training use. I also needed to develop a server that would execute these functionalities on command. I wanted to be able to design a full-stack system that allowed me to create every aspect of a user-facing server that can execute predictions using the classifiers that I design.
Throughout this project, I decided on a number of learning goals to consider it a success. I needed to learn how to use the supporting libraries that would help me to design this system. I also learned how to use the Twitter API, as well as create the infrastructure behind it that would allow me to collect large amounts of data for machine learning. I needed to become familiar with common machine learning libraries in Python in order to create the necessary algorithms and pipelines to make predictions based on Twitter data.
This paper details the steps and decisions needed to determine how to collect this data and apply it to machine learning algorithms. I determined how to create labelled data using pre-existing Botometer ratings, and the levels of confidence I needed to label data for training. I use the scikit-learn library to create these algorithms to best detect these bots. I used a number of pre-processing routines to refine the classifiers’ precision, including natural language processing and data analysis techniques. I eventually move to remotely-hosted versions of the system on Amazon web instances to collect larger amounts of data and train more advanced classifiers. This leads to the details of my final implementation of a user-facing server, hosted on AWS and interfacing over Gmail’s IMAP server.
The current and future development of this system is laid out. This includes more advanced classifiers, better data analysis, conversions to third party Twitter data collection systems, and user features. I detail what it is I have learned from this exercise, and what it is I hope to continue working on.