We demonstrate a computational phase correction algorithm that is able to correct for phase and timing fluctuations of arbitrary dual comb spectra. By augmenting a Kalman filter with a global search and decoupling the interferogram estimation, we show that dual comb signals having a wide range of structures can be predicted and corrected. Furthermore, we derive an upper bound for the accuracy of any self-correction technique and show that the augmented filter is capable of reaching this bound when the phase and frequency noise are bandlimited. Finally, we show how expectation maximization can be used to learn the statistical parameters of a system without any free parameters. This approach is hands-off, robust, and accurate for a wide range of dual comb systems. Demonstration code is provided.
Traditional recommendation systems are subject to a strong feedback loop by learning from and reinforcing past user-item interactions, which in turn limits the discovery of novel user interests. To address this, we introduce a hybrid hierarchical framework combining Large Language Models (LLMs) and classic recommendation models for user interest exploration. The framework controls the interfacing between the LLMs and the classic recommendation models through “interest clusters”, the granularity of which can be explicitly determined by algorithm designers. It recommends the next novel interests by first representing “interest clusters” using language, and employs a fine-tuned LLM to generate novel interest descriptions that are strictly within these predefined clusters. At the low level, it grounds these generated interests to an item-level policy by restricting classic recommendation models, in this case a transformer-based sequence recommender to return items that fall within the novel clusters generated at the high level. We showcase the efficacy of this approach on an industrial-scale commercial platform serving billions of users. Live experiments show a significant increase in both exploration of novel interests and overall user enjoyment of the platform.
1 March 2020 in Computational Statistics & Data Analysis
Precise knowledge of a laser’s wavelength is crucial for applications from spectroscopy to telecommunications. Here, we present a wavemeter that operates on the Talbot effect. Tone parameter extraction algorithms are used to retrieve the frequency of the periodic signal obtained in the sampled Talbot interferogram. Theoretical performance analysis based on the Cramér–Rao lower bound as well as experimental results are presented and discussed. With this scheme, we experimentally demonstrate a compact and high-precision wavemeter with below 10 pm single-shot estimation uncertainty under the criterion around 780 nm.