Total: 17
In supervised machine learning, model performance can decrease significantly when the distribution generating the new data varies from the distribution that generated the training data. One of the situations is covariate shift which happens a lot when labeled training data is missing, hard to get access to or very expensive to uniformly collect. All (probabilistic) classifiers will suffer from covariate shift. This motivates our research. Generally, we try to answer this question: how can we deal with covariate shift and generate predictions that are robust and reliable? We propose to develop a general framework for classification under covariate shift that is robust, flexible and accurate.
As natural language capable robots and other agents become more commonplace, the ability for these agents to understand truly natural human speech is becoming increasingly important. What is more, these agents must be able to understand truly natural human speech in realistic scenarios, in which an agent may not have full certainty in its knowledge of its environment, and in which an agent may not have full knowledge of the entities contained in its environment. As such, I am interested in developing architectural mechanisms which will allow robots to understand natural language in uncertain and open-worlds. My work towards this goal has primarily focused on two problems: (1) reference resolution, and (2) pragmatic reasoning.
Large state spaces and the curse of dimensionality contribute to the complexity of a task. Learning from demonstration techniques can be combined with reinforcement learning to narrow the exploration space of an agent, but require consistent and accurate demonstrations, as well as the state-action pairs for an entire demonstration. Individuals with severe motor disabilities are often slow and prone to human errors in demonstrations while teaching. My dissertation develops tools to allow persons with severe motor disabilities, and individuals in general, to train these systems. To handle these large state spaces as well as human error, we developed Dimensionality Reduced Reinforcement Learning. To accommodate slower feedback, we will develop a movie-reel style learning from demonstration interface.
Images can convey rich semantics and evoke strong emotions in viewers. The research of my PhD thesis focuses on image emotion computing (IEC), which aims to predict the emotion perceptions of given images. The development of IEC is greatly constrained by two main challenges: affective gap and subjective evaluation. Previous works mainly focused on finding features that can express emotions better to bridge the affective gap, such as elements-of-art based features and shape features. According to the emotion representation models, including categorical emotion states (CES) and dimensional emotion space (DES), three different tasks are traditionally performed on IEC: affective image classification, regression and retrieval. The state-of-the-art methods on the three above tasks are image-centric, focusing on the dominant emotions for the majority of viewers. For my PhD thesis, I plan to answer the following questions: (1) Compared to the low-level elements-of-art based features, can we find some higher level features that are more interpretable and have stronger link to emotions? (2) Are the emotions that are evoked in viewers by an image subjective and different? If they are, how can we tackle the user-centric emotion prediction? (3) For image-centric emotion computing, can we predict the emotion distribution instead of the dominant emotion category?
Effective communication through language involves organizing the content a person or system wishes to convey into text that flows naturally. There are many ways to render the same information, but those appropriate for one group of audience may not be intelligible to another. The goal of this thesis to analyze and address factors that influence the intelligibility of text from two aspects of information packaging: discourse structure and text specificity. Effective communication through language involves organizing the content a person or system wishes to convey into text that flows naturally. There are many ways to render the same information, but those appropriate for one group of audience may not be intelligible to another. The goal of this thesis to analyze and address factors that influence the intelligibility of text from two aspects of information packaging: discourse structure and text specificity.
This thesis explores the use of a recurrent neural network model for a novel story generation task. In this task, the model analyzes an ongoing story and generates a sentence that continues the story.
In online social networks (OSNs), users are allowed to create and share content about themselves and others. When multiple entities start distributing content, information can reach unintended individuals and inference can reveal more information about the user. Existing applications do not focus on detecting privacy violations before they occur in the system. This thesis proposes an agent-based representation of a social network, where the agents manage users' privacy requirements and create privacy agreements with agents. The privacy context, such as the relations among users, various content types in the system, and so on are represented with a formal language. By reasoning with this formal language, an agent checks the current state of the system to resolve privacy violations before they occur. We argue that commonsense reasoning could be useful to solve some of privacy examples reported in the literature. We will develop new methods to automatically identify private information using commonsense reasoning, which has never been applied to privacy context. Moreover, agents may have conflicting privacy requirements. We will study how to use agreement technologies in privacy settings for agents to resolve conflicts automatically.
Resource optimization and scheduling is a costly, challenging problem that affects almost every aspect of our lives. One example that affects each of us is health care: Poor systems design and scheduling of resources can lead to higher rates of patient noncompliance and burnout of health care providers, as highlighted by the Institute of Medicine (Brandenburg et al. 2015). In aerospace manufacturing, every minute re-scheduling in response to dynamic disruptions in the build process of a Boeing 747 can cost up to $100.000. The military is also highly invested in the effective use of resources. In missile defense, for example, operators must =solve a challenging weapon-to-target problem, balancing the cost of expendable, defensive weapons while hedging against uncertainty in adversaries’ tactics. Researchers in artificial intelligence (AI) planning and scheduling strive to develop algorithms to improve resource allocation. However, there are two primary challenges. First, optimal task allocation and sequencing with upper and lower-bound temporal constraints (i.e., deadlines and wait constraints) is NP-Hard (Bertsimas and Weismantel 2005). Approximation techniques for scheduling exist and typically rely on the algorithm designer crafting heuristics based on domain expertise to decompose or structure the scheduling problem and prioritize the manner in which resources are allocated and tasks are sequenced (Tang and Parker 2005; Jones, Dias, and Stentz 2011). The second problem is this aforementioned reliance on crafting clever heuristics based on domain knowledge. Manually capturing domain knowledge within a scheduling algorithm remains a challenging process and leaves much to be desired (Ryan et al. 2013). The aim of my thesis is to develop an autonomous system that 1) learns the heuristics and implicit rules-of-thumb developed by domain experts from years of experience, 2) embeds and leverages this knowledge within a scalable resource optimization framework, and 3) provides decision support in a way that engages users and benefits them in their decision-making process. By intelligently leveraging the ability of humans to learn heuristics and the speed of modern computation, we can improve the ability to coordinate resources in these time and safety-critical domains.
The goal of the research is to discover and summarize data from the emerging social media into information of interests. Specifically, leveraging user-contributed data from cross-domain social media, the idea is to perform multi-modal learning for a given photo, aiming to present people’s description or comments, geographical information, and events of interest, closely related to the photo. These information then can be used for various purposes, such as being a real-time guide for the tourists to improve the quality of tourism. As a result, this research investigates modern challenges of image annotation, image retrieval, and cross-media mining, followed by presenting promising ways to conquer the challenges.
Human language is extraordinarily creative in form and function, and adapting to this ever-shifting linguistic landscape is a daunting task for interactive cognitive systems. Recently, construction grammar has emerged as a linguistic theory for representing these complex and often idiomatic linguistic forms. Furthermore, analogical generalization has been proposed as a learning mechanism for extracting linguistic constructions from input. I propose an account that uses a computational model of analogy to learn and generalize argument structure constructions.
In many real-world domains, the presence of machines is becoming more ubiquitous to the point that they are usually more than simple automation tools. As part of the environment amongst human users, it is necessary for these computers and robots to be able to interact with them reasonably by either working independently around them or participating in a task, especially one with which a person needs help. This interactive procedure requires several steps: recognizing the user and environment from sensor data, interpreting the user’s activity and motives, determining a responsive behavior, performing the behavior, and then recognizing everything again to confirm the behavior choice and replan if necessary. At the moment, the research areas addressing these steps, activity recognition, plan recognition, intent recognition, and planning, have all been primarily studied independently. However, pipelining each independent process can be risky in real-time situations where there may be enough time to only run a few steps. This leads to a critical question: how do we perform everything under time constraints? In this thesis summary, I propose a framework that integrates these processes by taking advantage of features shared between them.
My thesis addresses the problem of planning under teammate behavior uncertainty by introducing the concept of intentional multiagent communication within ad hoc teams. In partially observable multiagent domains, agents much share information regarding aspects of the environment such that uncertainty is reduced across the team, permitting better coordination. Similarly, we consider how communication may be utilized within ad hoc teams to resolve behavioral uncertainty. Transmitting intentional messages allows agents to adjust predictions of a teammate's individual course of action. In short, an ad hoc agent coordinating with an unknown teammate can identify uncertainties within its own predictive model of teammate behavior then request the appropriate policy information, allowing the agent to adapt its personal plan. The main contribution of this work is the characterization of the interaction between learning, communication, and planning in ad hoc teams.
Autonomous systems must consider the moral ramifications of their actions. Moral norms vary among people and depend on common sense, posing a challenge for encoding them explicitly in a system. I propose to develop a model of repeated analogical chaining and analogical reasoning to enable autonomous agents to interactively learn to apply common sense and model an individual’s moral norms.
While Hierarchical Task Networks are frequently cited as flexible and powerful planning models, they are often ignored due to the intensive labor cost for experts/programmers, due to the need to create and refine the model by hand. While recent work has begun to address this issue by working towards learning aspects of an HTN model from demonstration, or even the whole framework, the focus so far has been on simple domains, which lack many of the challenges faced in the real world such as imperfect information and real-time environments. I plan to extend this work using the domain of real-time strategy (RTS) games, which have gained recent popularity as a challenging and complex domain for AI research.
Knowledge Graphs with rich schemas can allow for complex querying. My thesis focuses on providing accessible Knowledge using Gricean notions of Cooperative Answering as a motivation. More specifically, using Query Reformulations, Data Awareness, and a Pragmatic Context, along with the results they can become more responsive to user requirements and user context.
Markov Logic Networks (MLNs) use a few weighted first-order logic formulas to represent large probabilistic graphical models and are ideally suited for representing both relational and probabilistic knowledge in a wide variety of application domains such as, NLP, computer vision, and robotics. However, inference in them is hard because the graphical models can be extremely large, having millions of variables and features (potentials). Therefore, several lifted inference algorithms that exploit relational structure and operate at the compact first-order level, have been developed in recent years. However, the focus of much of existing research on lifted inference is on marginal inference while algorithms for MAP and marginal MAP inference are far less advanced. The aim of the proposed thesis is to fill this void, by developing next generation inference algorithms for MAP and marginal MAP inference.
Advances in sensing and imaging have provided psychology researchers new tools to understand how the brain creates the mind and simultaneously revealed the need for a new paradigm of mind-brain correspondence-- a set of basic theoretical tenets and an overhauled methodology. I develop machine learning methods to overcome three initial technical barriers to application of the new paradigm. I assess candidate solutions to these problems using two test datasets representing different areas of psychology: the first aiming to build more objective Post-Traumatic Stress Disorder(PTSD) diagnostic tools using virtual reality and peripheral physiology, the second aiming to verify theoretical tenets of the new paradigm in a study of basic affect using functional Magnetic Resonance Imaging(fMRI). Specifically I address three technical challenges: assessing performance in small, real datasets through stability; learning from labels of varying quality; and probabilistic representations of dynamical systems.