Noise-resilient and high-speed deep learning with coherent silicon photonics.

The explosive growth of deep learning applications has triggered a new era in computing hardware, targeting the efficient deployment of multiply-and-accumulate operations. In this realm, integrated photonics have come to the foreground as a promising energy efficient deep learning technology platform for enabling ultra-high compute rates. However, despite integrated photonic neural network layouts have already penetrated successfully the deep learning era, their compute rate and noise-related characteristics are still far beyond their promise for high-speed photonic engines. Herein, we demonstrate experimentally a noise-resilient deep learning coherent photonic neural network layout that operates at 10GMAC/sec/axon compute rates and follows a noise-resilient training model. The coherent photonic neural network has been fabricated as a silicon photonic chip and its MNIST classification performance was experimentally evaluated to support accuracy values of >99% and >98% at 5 and 10GMAC/sec/axon, respectively, offering 6× higher on-chip compute rates and >7% accuracy improvement over state-of-the-art coherent implementations.

A randomized, double-blind, phase II, exploratory trial evaluating the palliative benefit of either continuing pamidronate or switching to zoledronic acid in patients with high-risk bone metastases from breast cancer.

Previous studies suggest switching from pamidronate to a more potent bone-targeted agent is associated with biomarker and palliative response in breast cancer patients with bone metastases. Until now, this has not been addressed in a double-blind, randomized trial. Breast cancer patients with high-risk bone metastases, despite >3 months of pamidronate, were randomized to either continue pamidronate or switch to zoledronic acid every 4 weeks for 12 weeks. Primary outcome was the proportion of patients achieving a fall in serum C-telopeptide (sCTx) at 12 weeks. Secondary outcomes included difference in mean sCTx, pain scores, quality of life, toxicity, and skeletal-related events (SREs). Seventy-three patients entered the study; median age 61 years (range 37-87). Proportion of patients achieving a fall in sCTx over the 12-week evaluation period was 26/32 (81 %) with zoledronic acid and 18/29 (62 %) with pamidronate (p = 0.095). Mean decrease in sCTx (mean difference between groups = 50 ng/L, 95 % CI 18-84; p = 0.003) was significantly greater in patients who received zoledronic acid. Quality of life, pain scores, toxicity, and frequency of new SREs were comparable between the two arms. While a switch from pamidronate to zoledronic acid resulted in reduction in mean sCTx, there were no significant differences between the arms for proportion of patients achieving a reduction in sCTx, quality of life, pain scores, toxicity or SREs. Given the lack of palliative improvement, the current data do not support a switching strategy.

Effect of taurine on the concentrations of glutamate, GABA, glutamine and alanine in the rat striatum and hippocampus.

Taurine, a non-protein amino acid, acts as an osmoregulator and inhibitory neuromodulator in the brain. Here we studied the effects of intraperitoneal injections of taurine on the concentrations of glutamate and GABA, and their precursors, glutamine and alanine, in the rat striatum and hippocampus. Injections of 0.25, 0.5 and 1 g/kg taurine led to a gradual increase in taurine tissue concentrations in both hippocampus and striatum. Glutamate and GABA also increased in the hippocampus, but not in the striatum. Glutamine increased and alanine decreased markedly in both brain structures. The results corroborate the neuromodulatory role of taurine in the brain. Taurine administration results in an imbalance in inhibitory and excitatory neurotransmission in the glutamatergic (hippocampus) and GABAergic (striatum) brain structures, affecting more markedly the neurotransmitter precursors.