Filetype
Video data and the tools for automated analysis have a great potential to be used in road traffic research, particularly road safety. In this project a video dataset is built and made public so that researchers can evaluate their algorithms on it. The dataset focuses on the traffic research applications (data from real research projects) and provides recordings of the traffic scenes, meta-data, ca
In this paper, we describe how we use a named entity disambiguation module to merge entities in a question answering system. The question answering system uses a baseline passage retrieval component that extracts paragraphs from the Swedish version of Wikipedia. The passages are indexed and ranked using the Lucene platform. Prior to the indexing, we carried out a recognition and disambiguation of
In this paper we discuss AI in industrial robotics. In automatic control, computer vision and optimization, ma- chine learning and data mining algorithms are widely used. However, cognition enabling mechanisms, such as high-level logic and symbolic reasoning, are still limited. This is not due to the lack of available algorithms, rather the bottleneck is knowledge representation, acquisition and t
This thesis explores methods for generating proposition databases in a large-scale and multilingual setting. Our methods are centered on using semantic role labeling for extracting predicate-argument structures, and the subsequent transformation of such structures for knowledge base population and generation. By extending semantic role labeling with entity detection, we demonstrate how predicate-a
A smartphone is a personal device and as such usually hosts multiple public user identities such as a phone number, email address, and Facebook account. As each smartphone has a unique Bluetooth MAC address, Bluetooth discovery can be used in combination with the user registration to a server with a Facebook account. This makes it possible to identify a nearby smartphone related to a given Faceboo
Radio based localization and tracking usually require multiple receivers/transmitters or a known floor plan. This paper presents a method for anchor free indoor positioning based on single antenna ultra wideband (UWB) measurements. By using time of arrival information from multipath propagation components stemming from scatterers with different, but unknown, positions we estimate the movement of t
In this paper, we describe a passage retrieval component for a questioning answering system and we evaluate its performance on Swedish documents. We used a corpus of questions and answers transcribed from the Swedish board game Kvitt eller dubbelt and, as source for the passages, we used the articles of the Swedish version of Wikipedia. We show that Wikipedia is a suitable knowledge source to answ
Semantic role labeling has become a key module of many language processing applications. To build an unrestricted semantic role labeler, the first step is to develop a comprehensive proposition bank. However, building such a bank is a costly enterprise, which has only been achieved for a handful of languages. In this paper, we describe a technique to build proposition banks for new languages using
High resolution radio based positioning and tracking is a key enabler for new or improved cellular services. In this work, we are aiming to track user movements with accuracy down to centimeters using standard cellular bandwidths of 20-40 MHz. The goal is achieved by using phase information from the multi-path components (MPCs) of the radio channels. First, an extended Kalman filter (EKF) is used
In this paper, we describe KOSHIK, an end-to-end framework to process the unstructured natural language content of multilingual documents. We used the Hadoop distributed computing infrastructure to build this framework as it enables KOSHIK to easily scale by adding inexpensive commodity hardware. We designed an annotation model that allows the processing algorithms to incrementally add layers of a
Most question answering systems feature a step to predict an expected answer type given a question. Li and Roth \cite{li2002learning} proposed an oft-cited taxonomy to the categorize the answer types as well as an annotated data set. While offering a framework compatible with supervised learning, this method builds on a fixed and rigid model that has to be updated when the question-answering domai
The massive amounts of text data made available through the Google Books digitization project have inspired a new field of big-data textual research. Named culturomics, this field has attracted the attention of a growing number of scholars over recent years. However, initial studies based on these data have been criticized for not referring to relevant work in linguistics and language technology.
Things such as organizations, persons, or locations are ubiquitous in all texts circulating on the internet, particularly in the news, forum posts, and social media. Today, there is more written material than any single person can read through during a typical lifespan. Automatic systems can help us amplify our abilities to find relevant information, where, ideally, a system would learn knowledge
LSTM architectures (Hochreiter and Schmidhuber, 1997) have become standard to recognize named entities (NER) in text (Lample et al., 2016; Chiu and Nichols, 2016). Nonetheless, Zhang et al. (2015) recently proposed an approach based on fixed-size ordinally forgetting encoding (FOFE) to translate variable-length contexts into fixed-length features. This encoding method can be used with feed-forward
The role of time-dependent freezing of ice nucleating particles (INPs) is evaluated with the “Aerosol–Cloud” (AC) model in 1) deep convection observed over Oklahoma during the Midlatitude Continental Convective Cloud Experiment (MC3E), 2) orographic clouds observed over North California during the Atmospheric Radiation Measurement (ARM) Cloud Aerosol Precipitation Experiment (ACAPEX), and 3) super
To train well-performing generalizing neural networks, sufficiently large and diverse datasets are needed. Collecting data while adhering to privacy legislation becomes increasingly difficult and annotating these large datasets is both a resource-heavy and time-consuming task. An approach to overcome these difficulties is to use synthetic data since it is inherently scalable and can be automatical
Task-level industrial robot programming is a mundane, error-prone activity requiring expertise and skill. Since humans easily communicate with natural language (NL), it may be attractive to use speech or text as instruction means for robots. However, there has to be a substantial amount of knowledge in the system to translate the high-level language instructions to executable robot programs. In th
Named entity linking is the task of identifying mentions of named things in text, such as “Barack Obama” or “New York”, and linking these mentions to unique identifiers. In this paper, we describe Hedwig, an end-to-end named entity linker, which uses a combination of word and character BILSTM models for mention detection, a Wikidata and Wikipedia-derived knowledge base with global information aggr
We consider sequences of points obtained by projecting a given point B=B (0) back and forth between two manifolds and , and give conditions guaranteeing that the sequence converges to a limit . Our motivation is the study of algorithms based on finding the limit of such sequences, which have proved useful in a number of areas. The intersection is typically a set with desirable properties but for w
When robots are working in dynamic environments, close to humans lacking extensive knowledge of robotics, there is a strong need to simplify the user interaction and make the system execute as autonomously as possible, as long as it is feasible. For industrial robots working side-by-side with humans in manufacturing industry, AI systems are necessary to lower the demand on programming time and sys
