Apple has announced that it will be attending the 33rd Conference and Workshop on Neural Information Processing Systems (NeurIPS) in Vancouver, Canada from Sunday, December 8 through Saturday, December 14. In a new entry to its Machine Learning Journal, Apple said its product teams are "engaged in state of the art research in machine hearing, speech recognition, natural language processing, machine translation, text-to-speech, and artificial intelligence, improving the lives of millions of customers every day." Apple employees will be making a series of presentations at the conference. A schedule is provided in Apple's Machine Learning Journal. Machine learning algorithms play a role in virtually every Apple product and service, ranging from Apple Maps and Apple News to Siri and the QuickType keyboard on iPhone and iPad.

When I have written about Apple's AI and Machine Learning initiatives in the past, the articles have usually centered around Siri and Voice Dictation. It's easy to put these things together because Siri is the most visible and user-centered Apple interface that involves AI. However, as we have seen this month, there is a lot more than meets the eye going on beneath the surface at Apple. Last week, Wired ran a story about a lunch talk given by Apple's leading AI expert, Rutland Salakhutdinov, for around 200 others in the field during the NIPS machine learning conference. The most interesting thing to come out of his presentation was fresh news of Apple's continued work in the field of self-driving cars.

But bahgawd, the technology Apple is pulling off really is falling squarely into the realm of magical. Anyone else just wowed by the amount of technology embedded into this new iPhone? Our phones are learning more about us then we ever knew Before.

The consumer technology giant debuted Wednesday a website which highlights the provider's various AI-related studies. Named the Apple Machine Learning Journal, the tech giant is pitching it as an easy method for people to read about the job of its own various engineers taking care of cutting edge AI techniques like profound learning.Last year, Apple announced that it is starting gates to AI and allowing investigators to release their work for peer review. Learning technologies to help build advanced products for millions of people round the world posted in the Machine Learning Journal, it'd be obvious that journal is not meant for regular consumers.

The crucial importance of metrics in machine learning algorithms has led to an increasing interest in optimizing distance and similarity functions, an area of research known as metric learning. When data consist of feature vectors, a large body of work has focused on learning a Mahalanobis distance. Less work has been devoted to metric learning from structured objects (such as strings or trees), most of it focusing on optimizing a notion of edit distance. We identify two important limitations of current metric learning approaches. First, they allow to improve the performance of local algorithms such as k-nearest neighbors, but metric learning for global algorithms (such as linear classifiers) has not been studied so far. Second, the question of the generalization ability of metric learning methods has been largely ignored. In this thesis, we propose theoretical and algorithmic contributions that address these limitations. Our first contribution is the derivation of a new kernel function built from learned edit probabilities. Our second contribution is a novel framework for learning string and tree edit similarities inspired by the recent theory of (e,g,t)-good similarity functions. Using uniform stability arguments, we establish theoretical guarantees for the learned similarity that give a bound on the generalization error of a linear classifier built from that similarity. In our third contribution, we extend these ideas to metric learning from feature vectors by proposing a bilinear similarity learning method that efficiently optimizes the (e,g,t)-goodness. Generalization guarantees are derived for our approach, highlighting that our method minimizes a tighter bound on the generalization error of the classifier. Our last contribution is a framework for establishing generalization bounds for a large class of existing metric learning algorithms based on a notion of algorithmic robustness.