we're only beginning this web site and I believed its a great chance to share everything we want in. Accidentaly, I was expected to review modern development in our team and this is actually the summary.
Very first the greatest news, our group has grew over the past year to five PhD and around 10 MSc students in device learning.
Today I would like to focus on part one and mention a few of our development in machine learning. Within the second part I will explain our IoT work. The key device learning subjects can be broken-in listed here categories:
I’ll begin with the most important success, the YodaQA giving answers to device. Its an open resource concern answering system. It implements state-of-art methods of information extraction and natural language understanding — to resolve human-phrased questions! You can look at the real time demo.
With YodaQA we've worked additionally on simpler Intelligent Assistants acting on a smaller sized few instructions. They make the most from less complicated algorithms locating the most similar solution for certain question. The phrase set similarity is yet another topic interesting. The algorithms enables solving not merely Answer Sentence Selection, but in addition other interesting issues, such as for instance Following Utterance Ranking, Semantic Textual Similarity, Paraphrase Identification, acknowledging Textual Entailment etc. We tested and developed some algorithms predicated on term embeddings and various architecture of Neural Networks.
To your NLP group belongs additionally the multinomial category algorithm. The use case we are testing is the products categorization to a hierarchical directory structure. Often the e-shops are categorizing services and products, including a “14 inch display screen notebook” under notebooks, computer systems, electronic devices an such like. This process is handled by humans as well as do mistakes, our algorithm find problematically classified entries or recommend correct group.