RESUMEN
Conifer shoots exhibit intricate geometries at an exceptionally detailed spatial scale. Describing the complete structure of a conifer shoot, which contributes to a radiation scattering pattern, has been difficult, and the previous respective components of radiative transfer models for conifer stands were rather coarse. This paper presents a dataset aimed at models and applications requiring detailed 3D representations of needle shoots. The data collection was conducted in the Järvselja RAdiation transfer Model Intercomparison (RAMI) pine stand in Estonia. The dataset includes 3-dimensional surface information on 10 shoots of two conifer species present in the stand (5 shoots per species) - Scots pine (Pinus sylvestris L.) and Norway spruce (Picea abies L. Karst.). The samples were collected on 26th July 2022, and subsequently blue light 3D photogrammetry scanning technique was used to obtain their high-resolution 3D point cloud representations. For each of these samples, the dataset comprises of a photo of the sampled shoot and its obtained 3-dimensional surface reconstruction. Scanned shoots may replace previous, artificially generated models and contribute to the more realistic representation of 3D forest representations and, consequently, more accurate estimates of related parameters and processes by radiative transfer models.
RESUMEN
Deep learning applications require global optimization of non-convex objective functions, which have multiple local minima. The same problem is often found in physical simulations and may be resolved by the methods of Langevin dynamics with Simulated Annealing, which is a well-established approach for minimization of many-particle potentials. This analogy provides useful insights for non-convex stochastic optimization in machine learning. Here we find that integration of the discretized Langevin equation gives a coordinate updating rule equivalent to the famous Momentum optimization algorithm. As a main result, we show that a gradual decrease of the momentum coefficient from the initial value close to unity until zero is equivalent to application of Simulated Annealing or slow cooling, in physical terms. Making use of this novel approach, we propose CoolMomentum-a new stochastic optimization method. Applying Coolmomentum to optimization of Resnet-20 on Cifar-10 dataset and Efficientnet-B0 on Imagenet, we demonstrate that it is able to achieve high accuracies.
