Eric Yeh

WikiWalk: Random walks on Wikipedia for Semantic Relatedness

Computing semantic relatedness of natural language texts is a key component of tasks such as information retrieval and summarization, and often depends on knowledge of a broad range of real-world concepts and relationships. We address this knowledge integration issue by computing semantic relatedness using personalized PageRank (random walks) on a graph derived from Wikipedia. This paper evaluates methods for building the graph, including link selection strategies, and two methods for representing input texts as distributions over the graph nodes: one based on a dictionary lookup, the other based on Explicit Semantic Analysis. We evaluate our techniques on standard word relatedness and text similarity datasets, finding that they capture similarity information complementary to existing Wikipedia-based relatedness measures, resulting in small improvements on a state-of-the-art measure.

Learning Alignments and Leveraging Natural Logic

We describe an approach to textual inference that improves alignments at both the typed dependency level and at a deeper semantic level. We present a machine learning approach to alignment scoring, a stochastic search procedure, and a new tool that finds deeper semantic alignments, allowing rapid development of semantic features over the aligned graphs. Further, we describe a complementary semantic component based on natural logic, which shows an added gain of 3.13% accuracy on the RTE3 test set.

Language analytics for assessing brain health: cognitive impairment, depression and pre-symptomatic Alzheimer's disease

We present data demonstrating how brain health may be assessed by applying data-mining and text analytics to patient language. Three brain-based disorders are investigated - Alzheimers Disease, cognitive impairment and clinical depression. Prior studies identify particular language characteristics associated with these disorders. Our data show computer-based pattern recognition can distinguish language samples from individuals with and without these conditions. Binary classification accuracies range from 73% to 97% depending on details of the classification task. Text classification accuracy is known to improve substantially as training data approaches web-scale. Such a web scale dataset seems inevitable given the ubiquity of social computing and its language intensive nature. Given this context, we claim that the classification accuracy levels obtained in our experiments are significant findings for the fields of web intelligence and applied brain informatics.

Reducing Human Fatigue in Interactive Evolutionary Computation Through Fuzzy Systems and Machine Learning Systems

We describe two approaches to reducing human fatigue in interactive evolutionary computation (IEC). A predictor function is used to estimate the human users score, thus reducing the amount of effort required by the human user during the evolution process. The fuzzy system and four machine learning classifier algorithms are presented. Their performance in a real-world application, the IEC-based design of a micromachine resonating mass, is evaluated. The fuzzy system was composed of four simple rules, but was able to accurately predict the users score 77% of the time on average. This is equivalent to a 51 % reduction of human effort compared to using IEC without the predictor. The four machine learning approaches tested were k-nearest neighbors, decision tree, AdaBoosted decision tree, and support vector machines. These approaches achieved good accuracy on validation tests, but because of the great diversity in user scoring behavior, were unable to achieve equivalent results on the user test data.

ISOMER: Informative Segment Observations for Multimedia Event Recounting

This paper describes a system for multimedia event detection and recounting. The goal is to detect a high level event class in unconstrained web videos and generate event oriented summarization for display to users. For this purpose, we detect informative segments and collect observations for them, leading to our ISOMER system. We combine a large collection of both low level and semantic level visual and audio features for event detection. For event recounting, we propose a novel approach to identify event oriented discriminative video segments and their descriptions with a linear SVM event classifier. User friendly concepts including objects, actions, scenes, speech and optical character recognition are used in generating descriptions. We also develop several mapping and filtering strategies to cope with noisy concept detectors. Our system performed competitively in the TRECVID 2013 Multimedia Event Detection task with near 100,000 videos and was the highest performer in TRECVID 2013 Multimedia Event Recounting task.

Template-Based Structured Argumentation

A semiautomated approach to evidential reasoning uses template-based structured argumentation. A template captures best analytic practice as a hierarchically structured set of coordinated questions; an argument answers the questions posed by a template, including references to the source material used as evidence to support those answers. Graphical depictions of arguments readily convey lines of reasoning, from evidence through to conclusions, making it easy to compare and contrast alternative lines of reasoning. Collaborative analysis is supported via simultaneous access to arguments through web browser clients connected to a common argument server. This approach to analysis has been applied to a wide range of analytic problems and has been experimentally shown to speed the development and improve the quality of analytic assessments.

Depression and self-focused language in structured interviews with older men

The association between depression and self-focused language has been found to varying extents across studies. The presence or absence of the association may depend on the communicative context. Based on Becks depression model, a broad, evaluative self-focused question was predicted more likely to elicit a stronger association than a full interview containing a more heterogeneous question set of items. The spontaneous speech obtained during structured interviews of 26 depressed and nondepressed older men, an as-yet little studied population, was analyzed. Results were consistent with the hypothesis that association between self-focused language and depression was demonstrated in the target question but not across the entire interview. The results may explain some of the aforementioned discrepancies in prior studies.

Human-aided multi-sensor fusion

This paper discusses some fundamental requirements for a human-aided multi-sensor fusion system, and proposes an approach that implements such a system. This approach involves integration of the probabilistic argumentation system and the structural evidential argumentation system, which both are variants of the Dempster-Shafer belief function theory. An example is shown that illustrates how this integrated approach can be applied to missile defense applications.

Learning to ask the right questions to help a learner learn

Intelligent systems require substantial bodies of problem-solving knowledge. Machine learning techniques hold much appeal for acquiring such knowledge but typically require extensive amounts of user-supplied training data. Alternatively, informed question asking can supplement machine learning by directly eliciting critical knowledge from a user. Question asking can reduce the amount of training data required, and hence the burden on the user; furthermore, focused question asking holds significant promise for faster and more accurate acquisition of knowledge. In previous work, we developed static strategies for question asking that provide background knowledge for a base learner, enabling the learner to make useful generalizations even with few training examples. Here, we extend that work with a learning approach for automatically acquiring question-asking strategies that better accommodate the interdependent nature of questions. We present experiments validating the approach and showing its usefulness for acquiring efficient, context-dependent question-asking strategies.

Late fusion and calibration for multimedia event detection using few examples

The state-of-the-art in example-based multimedia event detection (MED) rests on heterogeneous classifiers whose scores are typically combined in a late-fusion scheme. Recent studies on this topic have failed to reach a clear consensus as to whether machine learning techniques can outperform rule-based fusion schemes with varying amount of training data. In this paper, we present two parametric approaches to late fusion: a normalization scheme for arithmetic mean fusion (logistic averaging) and a fusion scheme based on logistic regression, and compare them to widely used rule-based fusion schemes. We also describe how logistic regression can be used to calibrate the fused detection scores to predict an optimal threshold given a detection prior and costs on errors. We discuss the advantages and shortcomings of each approach when the amount of positives available for training varies from 10 positives (10Ex) to 100 positives (100Ex). Experiments were run using video data from the NIST TRECVID MED 2013 evaluation and results were reported in terms of a ranking metric: the mean average precision (mAP) and R0, a cost-based metric introduced in TRECVID MED 2013.