No abstract

Sentence matching is widely used in various natural language tasks such as natural language inference, paraphrase identification, and question answering. For these tasks, understanding logical and semantic relationship between two sentences is required but it is yet challenging. Although attention mechanism is useful to capture the semantic relationship and to properly align the elements of two sentences, previous methods of attention mechanism simply use a summation operation which does not retain original features enough. Inspired by DenseNet, a densely connected convolutional network, we propose a densely-connected co-attentive recurrent neural network, each layer of which uses concatenated information of attentive features as well as hidden features of all the preceding recurrent layers. It enables preserving the original and the co-attentive feature information from the bottommost word embedding layer to the uppermost recurrent layer. To alleviate the problem of an ever-increasing size of feature vectors due to dense concatenation operations, we also propose to use an autoencoder after dense concatenation. We evaluate our proposed architecture on highly competitive benchmark datasets related to sentence matching. Experimental results show that our architecture, which retains recurrent and attentive features, achieves state-of-the-art performances for most of the tasks.

An elastic printed circuit board (E-PCB) is a conductive framework used for the facile assembly of system-level stretchable electronics. E-PCBs require elastic conductors that have high conductivity, high stretchability, tough adhesion to various components, and imperceptible resistance changes even under large strain. We present a liquid metal particle network (LMP Net ) assembled by applying an acoustic field to a solid-state insulating liquid metal particle composite as the elastic conductor. The LMP Net conductor satisfies all the aforementioned requirements and enables the fabrication of a multilayered high-density E-PCB, in which numerous electronic components are intimately integrated to create highly stretchable skin electronics. Furthermore, we could generate the LMP Net in various polymer matrices, including hydrogels, self-healing elastomers, and photoresists, thus showing their potential for use in soft electronics.

Optogenetics is a powerful technique that allows target-specific spatiotemporal manipulation of neuronal activity for dissection of neural circuits and therapeutic interventions. Recent advances in wireless optogenetics technologies have enabled investigation of brain circuits in more natural conditions by releasing animals from tethered optical fibers. However, current wireless implants, which are largely based on battery-powered or battery-free designs, still limit the full potential of in vivo optogenetics in freely moving animals by requiring intermittent battery replacement or a special, bulky wireless power transfer system for continuous device operation, respectively. To address these limitations, here we present a wirelessly rechargeable, fully implantable, soft optoelectronic system that can be remotely and selectively controlled using a smartphone. Combining advantageous features of both battery-powered and battery-free designs, this device system enables seamless full implantation into animals, reliable ubiquitous operation, and intervention-free wireless charging, all of which are desired for chronic in vivo optogenetics. Successful demonstration of the unique capabilities of this device in freely behaving rats forecasts its broad and practical utilities in various neuroscience research and clinical applications.