No abstract

No abstract

Artificial intelligence is facilitating human life in many aspects. Previous artificial intelligence has been mainly focused on computer algorithms (e.g. deep-learning and extremelearning) and integrated circuits. Recently, all-optical diffractive deep neural networks (D 2 NN) were realized by using passive structures, which can perform complicated functions designed by computer-based neural networks at the light speed. However, once a passive D 2 NN architecture is fabricated, its function will be fixed. Here, we propose a programmable artificial intelligence machine (PAIM) that can execute various intellectual tasks by realizing hierarchical connections of brain neurons via a multi-layer digital-coding metasurface array. Integrated with two amplifier chips in each meta-atom, its transmission coefficient covers a dynamic range of 35 dB (from -40 dB to -5 dB), which is the basis to construct the reprogrammable physical layers of D 2 NN, in which the digital meta-atoms make the artificial neurons alive. We experimentally show that PAIM can handle various deep-learning tasks for wave sensing, including image classifications, mobile communication coder-decoder, and real-time multi-beam focusing. In particular, we propose a reinforcement learning algorithm for on-site learning and discrete optimization algorithm for digital coding, making PAIM have autonomous intelligence ability and perform self-learning tasks without the support of extra computer.

Epithelial ovarian cancer (EOC) is one of the most common gynecological cancers, with diagnosis often at a late stage. Metastasis is a major cause of death in patients with EOC, but the underlying molecular mechanisms remain obscure. Here, we utilized an integrated approach to find potential key transcription factors involved in ovarian cancer metastasis and identified STAT4 as a critical player in ovarian cancer metastasis. We found that activated STAT4 was overexpressed in epithelial cells of ovarian cancer and STAT4 overexpression was associated with poor outcome of ovarian cancer patients, which promoted metastasis of ovarian cancer in both in vivo and in vitro. Although STAT4 mediated EOC metastasis via inducing epithelial-to-mesenchymal transition (EMT) of ovarian cancer cells in vivo, STAT4 failed to induce EMT directly in vitro, suggesting that STAT4 might mediate EMT process via cancer-stroma interactions. Further functional analysis revealed that STAT4 overexpression induced normal omental fibroblasts and adipose- and bone marrow-derived mesenchymal stem cells to obtain cancer-associated fibroblasts (CAF)-like features via induction of tumor-derived Wnt7a. Reciprocally, increased production of CAF-induced CXCL12, IL6 and VEGFA within tumor microenvironment could enable peritoneal metastasis of ovarian cancer via induction of EMT program. In summary, our study established a model that STAT4 promotes ovarian cancer metastasis via tumor-derived Wnt7a-induced activation of CAFs.

Electromagnetic (EM) waves have been widely applied in wireless communications, radar detection, navigation, and target recognition. Radiation and scattering are two common behaviors in the EM community, but it remains a long‐standing challenge to control them in a dynamical way, especially using a single, low‐cost, and compact hardware. Here, a promising solution is proposed by combining a programmable metasurface with a radiation array, which can manipulate the scattering properties, digitally and in real‐time, and exhibit different radiation modes simultaneously. More advantageous over previous investigations with the fixed radiation‐scattering performance, a field‐programmable gate array is introduced to extend, realize, and verify the multiple functions of the meta‐microstructure (MMS). As a proof‐of‐concept, multiple functions, including polarization conversion, scattering beam manipulation, diffusion scattering, radar cross‐section reduction, EM waves radiation, and vortex beam generation, have been adequately demonstrated by the MMS prototype.