Integrated Embedding Approach for Knowledge Base Completion with CNN

To tackle specific problems in knowledge base completion such as computational complexity and complex relations or nodes with high indegree or outdegree, an algorithm called IEAKBC(short for Integrated Embedding Approach for Knowledge Base Completion) is proposed, in which entities and relations from triplets are first mapped into low-dimensional vector spaces, each original triplet represented in the form of 3-column, k dimensional matrix; then features from different relations are integrated into head and tail entities thus forming fused triplet matrices used as another input channel for convolution. In CNN feature maps are extracted by filters, concatenated and weighted for output scores to discern whether the original triplet holds or not. Experiments show that IEAKBC holds certain advantages over other models; when scaling up to relatively larger datasets, signs of superiority of IEAKBC stand out especially on relations with high cardinalities. At last we apply IEAKBC to a personalized search application, comparing its performance with strong baselines to verify its practicality in real environments.

Output Feedback of Systems with Uncertain Delays, Parameters, and ODE State

This chapter focuses on output feedback of systems with uncertain delays, parameters, and ODE state. The previous chapter solved the problem of adaptive stabilization of multi-input LTI systems under the assumption that distinct discrete multiple actuator delays are long and unknown, and the actuator state in each input channel is unavailable for measurement. A local stabilization result is achieved. This chapter considers the more challenging case where uncertainties exist not only in the input delay and the PDE actuator state, but also appear in the ODE plant state and the parameters of the system and input matrices. To be specific, it addresses the uncertainty collections. Similar to the result in the previous chapter, when the actuator state is not measurable and the delay value is unknown at the same time, the result of multi-input delays obtained in this chapter is also not global.

A Two Microphone-Based Approach for Detecting and Identifying Speech Sounds in Hearing Support System

For people with hearing disabilities, not only would give them difficulties in going through their everyday lives but also sometimes could be life threatening. In this research we proposed a simple, yet robust approach for helping the hearing-impaired people in identifying the important sounds around them by using two microphones as input channel that could be worn around the person’s head as a substitute for their ears. This device then could be used to record the situation of the surroundings, and then the system would estimate the Direction of Arrival (DOA) of the sound sources, then detect and classify them using Support Vector Machine (SVM) into target speech or noise category. As the results, system’s classifier could produce FAR and FRR as low as 2%, in which 274 out of 280 samples were successfully classified as target speeches and 22 from the total of 27 noise samples were successfully classified as noise