June's NAACL conference saw machine learning specialists from technology company Iprova present a paper introducing a new and effective method for the unsupervised training of machine learning algorithms to infer sentence embeddings. The NAACL (North American Chapter of the Association for Computational Linguistics) Human Language Technologies (HLT) conference took place at the Hyatt Regency New Orleans hotel, Louisiana, from June 1–6, 2018. The research paper, entitled "Unsupervised Learning of Sentence Embeddings using Compositional n-Gram Features", will be presented by Matteo Pagliardini. Pagliardini is a senior machine learning engineer at Iprova and one of the three scientists that authored the research paper and developed the new model for unsupervised training, Sent2Vec. While there have been several successes in deep learning in recent years, the paper notes that these have almost exclusively relied on supervised training.
A law professor at the University of Surrey is arguing that it should be possible for computer-based artificial intelligence (AI) systems to be formally considered as inventors for any invention they contribute to, much in the same way a person would. The argument forms part of a paper, which has been published in the Boston College Law Review, entitled I Think, Therefore I Invent: Creative Computers and the Future of Patent Law. In its introduction the report makes the point that while inventions by computers have been granted patents previously, the concept of computer inventorship has never actually been considered by the courts. The concept of giving creative computers the credit for their own inventions may sound surreal but, in reality, they have been generating potentially patentable ideas for decades without acknowledgment. As Professor Ryan Abbott points out in his paper, 'machines have been autonomously generating patentable results for at least twenty years and the pace of such invention is likely increasing.'
To patent machine learning, you will need to correctly claim and describe your invention while making sure that you comply with current laws related to this type of intellectual property. Machine learning plays an important role in much of today's technology. For example, without machine learning, effective internet searches would not be possible. The problem with machine learning, however, is that it can be hard to file patents for inventions in this growing field. It is not always immediately clear what software inventions are eligible for patent protections.
So if monkeys and other animals can't own copyrights, can artificial intelligence create protected intellectual property? A new patent filing in the U.K. aims to find out. An international team led by AI activist Ryan Abbott, a law professor at the University of Surrey, has filed the first-ever patent applications for two inventions created autonomously by artificial intelligence without a human inventor. The AI inventor, named DABUS by its creator, Stephen Thaler, was previously best known for creating surreal art, but it was designed to come up with new ideas and then assess those ideas for consequences, novelty, and salience. So far the series of neural networks that makes up DABUS has come up with two ideas that may be worth patenting.
On March 15, DeepMind's AlphaGo, a computer powered by a self-learning artificial intelligence computer program, defeated Go grandmaster Lee Sedol. As the AI community celebrates this major milestone in making machines smart, the debate of "man vs. machine" is heating up. Over the past 25 years -- especially the last five years -- the AI community has transformed theoretical machine learning constructs to solve useful problems. AI techniques such as self-learning, reinforcement learning, and deep neural networks were developed to recognize traffic signs and classify images. The recent rapid progress in AI was powered by the dramatic increase in financial investments in AI.