Search for: All records

Physical reservoir computing leverages the intrinsic history-dependence and nonlinearity of hardware to encode spatiotemporal signals directly at the sensor level, enabling low-latency processing of dynamic inputs. Encoding delity depends on the separability of multi-state outputs, yet in practice it is often hampered by empirically chosen, suboptimal operating conditions. Here, we apply Bayesian optimization to improve the encoding performance of solution-processed Al₂O₃/In₂O₃ thin- lm transistors. By exploring a ve-dimensional pulse-parameter input space and using the normalized degree of separation for output state distinguishability, we demonstrate high- delity 6-bit temporal encoding corresponding to 64 output states. We further show that a model based on simpler 4-bit data can effectively guide optimization for more complex 6-bit tasks, substantially reducing experimental effort. Using a six-frame moving-car image sequence as a benchmark, we nd that the optimized 6-bit pulse conditions signi cantly enhance encoding accuracy, with 4-bit derived parameters performing comparably in terms of pixel errors. Shapley Additive Explanations (SHAP) analysis further reveals that gate-pulse amplitude and drain voltage are the dominant contributors to output state separation. This work establishes a data-driven strategy for identifying optimal operating conditions in reservoir devices and outlines a framework that can be transferred to diverse material platforms and physical reservoir implementations. 

Abstract The development of superconducting quantum processors relies on understanding and mitigating decoherence in superconducting qubits. Piezoelectric coupling contributes to decoherence by mediating energy exchange between microwave photons and acoustic phonons. Although bulk centrosymmetric materials like silicon and sapphire are non-piezoelectric and commonly used as qubit substrates, the lack of centrosymmetry at interfaces may induce piezoelectric losses. This effect was predicted decades ago but never experimentally observed in superconducting devices. Here, we report interface piezoelectricity at aluminum-silicon junctions and demonstrate it as a significant loss channel in superconducting devices. Using aluminum interdigital transducers on silicon, we observe piezoelectric transduction from room to millikelvin temperatures, with an effective electromechanical coupling factorK² ≈ (3 ± 0.4) × 10⁻⁵%, comparable to weakly piezoelectric substrates. Modeling shows this mechanism limits qubit quality factors toQ ~ 10⁴ − 10⁸, depending on surface participation and mode matching. These findings reveal interface piezoelectricity as a major dissipation channel and highlight the need for heterostructure and phononic engineering in next-generation superconducting qubits. 

Abstract We analyzed 19,123 natural language processing-related studies to explore the differences in task distributions and application contexts between large language models (LLMs) and non-LLM methods in health care. Through topic modeling analysis, we found that LLMs demonstrate advantages in open-ended tasks, while non-LLM methods dominate in information extraction tasks. These findings highlight the complementary strengths of the two technical paradigms and provide reference for their integration strategies in future health care applications. 

Abstract Percutaneous nephrostomy is widely used in kidney access surgeries. Despite its prevalence in urological interventions, it presents two operational challenges: 1) precise needle placement into the renal pelvis; and 2) avoiding hemorrhage from blood vessel rupture. In this study, we developed an endoscopic optical coherence tomography probe for needle navigation. We conducted experiments on thirty-one human kidneys for two aspects: 1) tissue recognition, and 2) blood vessel detection. Experimental results indicated that renal tissues including cortex, medulla, calyx, sinus fat, and pelvis could be effectively distinguished through structural optical coherence tomography imaging, and renal blood flow could be detected through the Doppler function. Deep learning methods were utilized to automate recognition procedures. For tissue classification, an Inception model was used, achieving a recognition accuracy of 99.6%. For blood vessel detection, an nnU-net model was applied, exhibiting an intersection over union value of 0.8917 for blood vessel and 0.9916 for background.