AI technology has made remarkable achievements in computational pathology (CPath), especially with the help of deep neural networks. However, the network performance is highly related to architecture design, which commonly requires human experts with
domain knowledge. In this paper, we combat this challenge with the recent advance in neural architecture search (NAS) to find an optimal network for CPath applications. In particular, we use differentiable architecture search (DARTS) for its efficiency. We first adopt a probing metric to show that the original DARTS lacks proper hyperparameter tuning on the CIFAR dataset, and how the generalization issue can be addressed using an adaptive optimization strategy. We then apply our searching framework on CPath applications by searching for the optimum network architecture on a histological tissue type dataset (ADP). Results show that the searched network outperforms state-of-the-art networks in terms of prediction accuracy and computation complexity. We further conduct extensive experiments to demonstrate the transferability of the searched network to new CPath applications, the robustness against downscaled inputs, as well as the reliability of predictions.
We will show the Cheeger-Colding segment inequality for manifolds with integral Ricci curvature bound. By using this segment inequality, the almost rigidity structure results for integral Ricci curvature will be derived by a similar method as in cite
{CC1}. And the sharp Holder continuity result of cite{CoN} holds in the limit space of manifolds with integral Ricci curvature bound.
We will show that the quantitative maximal volume entropy rigidity holds on Alexandrov spaces. More precisely, given $N, D$, there exists $epsilon(N, D)>0$, such that for $epsilon<epsilon(N, D)$, if $X$ is an $N$-dimensional Alexandrov space with cur
vature $geq -1$, $operatorname{diam}(X)leq D, h(X)geq N-1-epsilon$, then $X$ is Gromov-Hausdorff close to a hyperbolic manifold. This result extends the quantitive maximal volume entropy rigidity of cite{CRX} to Alexandrov spaces.
Spin waves (SWs), the collective precessional motion of spins in a magnetic system, have been proposed as a promising alternative system with low-power consumption for encoding information. Spin Hall nano-oscillator (SHNO), a new-type spintronic nano
-device, can electrically excite and control spin waves in both nanoscale magnetic metals and insulators with low damping by the spin current due to spin Hall effect. Here, we will review recent progress about spin-wave excitation and experimental parameters dependent spectrum in SHNOs. The nanogap SHNOs based on in-plane magnetization Py/Pt exhibits a nonlinear self-localized bullet soliton localized at the center of the gap between the electrodes and a secondary high-frequency mode which coexists with the primary bullet mode at higher currents. While in the nanogap SHNOs with strong perpendicular magnetic anisotropy (PMA), besides both nonlinear bullet soliton and propagating spin-wave mode are achieved and controlled by varying the external magnetic field and current, the magnetic bubble skyrmion mode also can be excited at a low in-plane magnetic field. These SW modes show thermal-induced mode hopping behavior at high temperature due to the coupling between modes mediated by thermal-magnon-mediated scattering. Moreover, thanks to PMA-induced effective field, a single coherent mode also can be achieved without applying an external magnetic field. The strong nonlinear effect of spin-waves makes SHNOs easy to achieve synchronization with external microwave signals or mutual synchronization between multiple oscillators with improving the coherence and power of oscillation modes significantly. Spin-waves in SHNOs with an external free magnetic layer have a wide range of applications from as a nanoscale signal source of low-power consumption magnonic devices to spin-based neuromorphic computing systems in the field of artificial intelligence.
Understanding and manipulating properties emerging at a surface or an interface require a thorough knowledge of structure-property relationships. We report a study of a prototype oxide system, La2/3Sr1/3MnO3 grown on SrTiO3(001), by combining in-situ
angle-resolved x-ray photoelectron spectroscopy, ex-situ x-ray diffraction, and scanning transmission electron microscopy/spectroscopy with electric transport measurements. We find that La2/3Sr1/3MnO3 films thicker than 20 unit cells (u.c.) exhibit a universal behavior with no more than one u.c. intermixing at the interface but at least 3 u.c. of Sr segregation near the surface which is (La/Sr)O terminated. The conductivity vs film thickness shows the existence of nonmetallic layers with thickness ~ 6.5 +/- 0.9 u.c., which is independent of film thickness but mainly relates to the deviation of Sr concentration near the surface region. Below 20 u.c., the surface of the films appears mixed (La/Sr)O with MnO2 termination. Decreasing film thickness to less than 10 u.c. leads to the enhanced deviation of chemical composition in the films and eventually drives the film insulating. Our observation offers a natural explanation for the thickness-driven metal-nonmetal transition in thin films based on the variation of film stoichiometry.
We experimentally study the dynamical modes excited by spin current in Spin Hall nano-oscillators based on the Pt/[Co/Ni] multilayers with perpendicular magnetic anisotropy. Both propagating spin wave and localized solitonic modes of the oscillation
are achieved and controlled by varying the applied magnetic field and current. At room temperature, the generation linewidth broadening associated with mode hopping was observed at currents close to the transition between different modes and in the mode coexistence regimes. The mode hopping was suppressed at cryogenic temperatures, confirming that the coupling between modes is mediated by thermal magnons. We also demonstrate that coherent single-mode oscillations with linewidth of 5 MHz can be achieved without applying external magnetic field. Our results provide insight into the mechanisms controlling the dynamical coherence in nanomagnetic oscillators, and guidance for optimizing their applications in spin wave-based electronics.
Spintronic nanodevices have ultrafast nonlinear dynamic and recurrence behaviors on a nanosecond scale that promises to enable spintronic reservoir computing (RC) system. Here two physical RC systems based on a single magnetic skyrmion memristor (MSM
) and 24 spin-torque nano-oscillators (STNOs) were proposed and modeled to process image classification task and nonlinear dynamic system prediction, respectively. Based on our micromagnetic simulation results on the nonlinear responses of MSM and STNO with current pulses stimulation, the handwritten digits recognition task domesticates that an RC system using one single MSM has the outstanding performance on image classification. In addition, the complex unknown nonlinear dynamic problems can also be well solved by a physical RC system consisted of 24 STNOs confirmed in a second-order nonlinear dynamic system and NARMA10 tasks. The capability of both high accuracy and fast information processing promises to enable one type of brain-like chip based on spintronics for various artificial intelligence tasks.
We give several Bishop-Gromov relative volume comparisons with integral Ricci curvature which improve the results in cite{PW1}. Using one of these volume comparisons, we derive an estimate for the volume entropy in terms of integral Ricci curvature w
hich substantially improves an earlier estimate in cite{Au2} and give an application on the algebraic entropy of its fundamental group. We also extend the almost minimal volume rigidity of cite{BBCG} to integral Ricci curvature.
The Milnor Problem (modified) in the theory of group growth asks whether any finite presented group of vanishing algebraic entropy has at most polynomial growth. We show that a positive answer to the Milnor Problem (modified) is equivalent to the Nil
potency Conjecture in Riemannian geometry: given $n, d>0$, there exists a constant $epsilon(n,d)>0$ such that if a compact Riemannian $n$-manifold $M$ satisfies that Ricci curvature $op{Ric}_Mge -(n-1)$, diameter $dge op{diam}(M)$ and volume entropy $h(M)<epsilon(n,d)$, then the fundamental group $pi_1(M)$ is virtually nilpotent. We will verify the Nilpotency Conjecture in some cases, and we will verify the vanishing gap phenomena for more cases i.e., if $h(M)<epsilon(n,d)$, then $h(M)=0$.
This is the second paper of two in a series under the same title ([CRX]); both study the quantitative volume space form rigidity conjecture: a closed $n$-manifold of Ricci curvature at least $(n-1)H$, $H=pm 1$ or $0$ is diffeomorphic to a $H$-space f
orm if for every ball of definite size on $M$, the lifting ball on the Riemannian universal covering space of the ball achieves an almost maximal volume, provided the diameter of $M$ is bounded for $H e 1$. In [CRX], we verified the conjecture for the case that $M$ or its Riemannian universal covering space $tilde M$ is not collapsed for $H=1$ or $H e 1$ respectively. In the present paper, we will verify this conjecture for the case that Ricci curvature is also bounded above, while the above non-collapsing condition is not required.
