A 1-by-2 compact injection-based switch is reported utilizing interior-ridge ring resonators, demonstrating nanosecond-scale switching speed, single-mode, low-loss (0/1.3dB for thru/drop port) with efficient integrated thermal tuner (7.25μW/GHz). OCIS codes: (130.4815) Optical switching devices; (130.3120) Integrated optics devices; (200.6715) Switching.

We integrate an efficient thermal tuner (6.2-!W/GHz) within ARM modulators for the first time, yielding 5-THz uncorrupted FSR. The 5.4-um diameter modulator achieves low-power (9-fJ/bit) and high-speed (13-Gb/s) operation with 5.9dB extinction ratio. OCIS codes: (130.3120) Integrated optics devices; (230.5750) Resonators; (230.7370) Waveguides

The primary reason for diversity of distributed storage systems is that no single design for a distributed storage system is cost-optimal for meeting the latency and throughput SLOs in all possible workloads. In other words, a distributed object store that satisfies the SLOs for a particular workload at minimum cost (in comparison to other object stores) will incur higher costs than alternate s...

Although many efforts have been exerted, implementation of fully integrated chip for TV tuner application is still a great challenge, to the standard of DVB-C (50M-860MHz) for example. Recent works on wideband LNA [1-2] based on off-chip balun have shown good performances, such as low NF, high linearity and excellent input matching. However, the off-chip balun adds extra cost, loss and much PCB...

Iterative gradient-based algorithms have been increasingly applied for the training of a broad variety machine learning models including large neural-nets. In particular, momentum-based methods, with accelerated guarantees, received lot attention due to their provable guarantees fast in certain classes problems and multiple derived. However, properties these methods hold only constant regressor...

An alternative approach for intelligent tuning of a control loop will be presented in this paper. The objective is to design an algorithm which will tune the controller employing a neuro-fuzzy based algorithm. Structure and design method based on this approach will be explored, which include an adaptive network employed as building block, the backpropagation gradient method and least square est...

One of our most beloved music teachers emphasized the importance of " facing the music, " by which he meant listening to recordings of our playing. As with the first hearing of one's voice on recording, many were both surprised and suspicious of this external perspective. While sometimes revealing more than we were ready to hear, the long-term effect of this exercise helps us to " hear ourselve...

Model induction is one of the most popular methods to extract information better understand AI’s decisions by estimating contribution input features for a class interest. However, we found potential issue: model methods, especially those that compute activation maps, rely on arbitrary thresholding mute some their computed attribution scores, which can cause severe quality degradation induction....