Language in a (Search) Box: Grounding Language Learning in Real-World Human-Machine Interaction

We investigate grounded language learning through real-world data, by modelling a teacher-learner dynamics through the natural interactions occurring between users and search engines; in particular, we explore the emergence of semantic generalization from unsupervised dense representations outside of synthetic environments. A grounding domain, a denotation function and a composition function are learned from user data only. We show how the resulting semantics for noun phrases exhibits compositional properties while being fully learnable without any explicit labelling. We benchmark our grounded semantics on compositionality and zero-shot inference tasks, and we show that it provides better results and better generalizations than SOTA non-grounded models, such as word2vec and BERT.

Human Interactive Method for a Ground Target Tracking Using Airborne Imaging System

This research presents a human interactive method for ground target tracking using airborne imaging system. This method depends on Template Matching using Zero Mean Normalized Cross Correlation.

Appearance based and Gesture Independent Model for Natural Human Hand Detection and Tracking

Many researchers work on enhancement of Human Computer Interaction methods and try to make it more natural and intuitive. This includes researches in: understanding human languages, gestures recognition and brain signals recognition. But the heavy use of hands in human everyday life makes hand recognition and tracking researches very important. In this paper, we present a novel method to recognize and track a human hand moving in front of digital camera in an unknown environment without any constraints on fingers positions or hand gesture and with no need to wear any additional devices like gloves or markers. Our method can distinguish between hand and other moving objects especially faces, by applying some proposed criteria to determine which object is representing the hand. A practical study is performed to evaluate the performance of the proposed method. Hand interactive virtual TV is made as a realistic application to report users experiences. Results show that our proposed method can recognize human hand at real-time with 99% accuracy rate in normal indoors light.