Evaluating Neural Radiance Fields for 3D Plant Geometry Reconstruction in Field Conditions.

We evaluate different Neural Radiance Field (NeRF) techniques for the 3D reconstruction of plants in varied environments, from indoor settings to outdoor fields. Traditional methods usually fail to capture the complex geometric details of plants, which is crucial for phenotyping and breeding studies. We evaluate the reconstruction fidelity of NeRFs in 3 scenarios with increasing complexity and compare the results with the point cloud obtained using light detection and ranging as ground truth. In the most realistic field scenario, the NeRF models achieve a 74.6% F1 score after 30 min of training on the graphics processing unit, highlighting the efficacy of NeRFs for 3D reconstruction in challenging environments. Additionally, we propose an early stopping technique for NeRF training that almost halves the training time while achieving only a reduction of 7.4% in the average F1 score. This optimization process substantially enhances the speed and efficiency of 3D reconstruction using NeRFs. Our findings demonstrate the potential of NeRFs in detailed and realistic 3D plant reconstruction and suggest practical approaches for enhancing the speed and efficiency of NeRFs in the 3D reconstruction process.

Out-of-Distribution Detection Algorithms for Robust Insect Classification.

Plants encounter a variety of beneficial and harmful insects during their growth cycle. Accurate identification (i.e., detecting insects' presence) and classification (i.e., determining the type or class) of these insect species is critical for implementing prompt and suitable mitigation strategies. Such timely actions carry substantial economic and environmental implications. Deep learning-based approaches have produced models with good insect classification accuracy. Researchers aim to implement identification and classification models in agriculture, facing challenges when input images markedly deviate from the training distribution (e.g., images like vehicles, humans, or a blurred image or insect class that is not yet trained on). Out-of-distribution (OOD) detection algorithms provide an exciting avenue to overcome these challenges as they ensure that a model abstains from making incorrect classification predictions on images that belong to non-insect and/or untrained insect classes. As far as we know, no prior in-depth exploration has been conducted on the role of the OOD detection algorithms in addressing agricultural issues. Here, we generate and evaluate the performance of state-of-the-art OOD algorithms on insect detection classifiers. These algorithms represent a diversity of methods for addressing an OOD problem. Specifically, we focus on extrusive algorithms, i.e., algorithms that wrap around a well-trained classifier without the need for additional co-training. We compared three OOD detection algorithms: (a) maximum softmax probability, which uses the softmax value as a confidence score; (b) Mahalanobis distance (MAH)-based algorithm, which uses a generative classification approach; and (c) energy-based algorithm, which maps the input data to a scalar value, called energy. We performed an extensive series of evaluations of these OOD algorithms across three performance axes: (a) Base model accuracy: How does the accuracy of the classifier impact OOD performance? (b) How does the level of dissimilarity to the domain impact OOD performance? (c) Data imbalance: How sensitive is OOD performance to the imbalance in per-class sample size? Evaluating OOD algorithms across these performance axes provides practical guidelines to ensure the robust performance of well-trained models in the wild, which is a key consideration for agricultural applications. Based on this analysis, we proposed the most effective OOD algorithm as wrapper for the insect classifier with highest accuracy. We presented the results of its OOD detection performance in the paper. Our results indicate that OOD detection algorithms can significantly enhance user trust in insect pest classification by abstaining classification under uncertain conditions.

Dominating Set Model Aggregation for communication-efficient decentralized deep learning.

Decentralized deep learning algorithms leverage peer-to-peer communication of model parameters and/or gradients over communication graphs among the learning agents with access to their private data sets. The majority of the studies in this area focus on achieving high accuracy, with many at the expense of increased communication overhead among the agents. However, large peer-to-peer communication overhead often becomes a practical challenge, especially in harsh environments such as for an underwater sensor network. In this paper, we aim to reduce communication overhead while achieving similar performance as the state-of-the-art algorithms. To achieve this, we use the concept of Minimum Connected Dominating Set from graph theory that is applied in ad hoc wireless networks to address communication overhead issues. Specifically, we propose a new decentralized deep learning algorithm called minimum connected Dominating Set Model Aggregation (DSMA). We investigate the efficacy of our method for different communication graph topologies with a small to large number of agents using varied neural network model architectures. Empirical results on benchmark data sets show a significant (up to 100X) reduction in communication time while preserving the accuracy or in some cases, increasing it compared to the state-of-the-art methods. We also present an analysis to show the convergence of our proposed algorithm.

On Consensus-Optimality Trade-offs in Collaborative Deep Learning.

In distributed machine learning, where agents collaboratively learn from diverse private data sets, there is a fundamental tension between consensus and optimality. In this paper, we build on recent algorithmic progresses in distributed deep learning to explore various consensus-optimality trade-offs over a fixed communication topology. First, we propose the incremental consensus-based distributed stochastic gradient descent (i-CDSGD) algorithm, which involves multiple consensus steps (where each agent communicates information with its neighbors) within each SGD iteration. Second, we propose the generalized consensus-based distributed SGD (g-CDSGD) algorithm that enables us to navigate the full spectrum from complete consensus (all agents agree) to complete disagreement (each agent converges to individual model parameters). We analytically establish convergence of the proposed algorithms for strongly convex and nonconvex objective functions; we also analyze the momentum variants of the algorithms for the strongly convex case. We support our algorithms via numerical experiments, and demonstrate significant improvements over existing methods for collaborative deep learning.

A fast saddle-point dynamical system approach to robust deep learning.

Recent focus on robustness to adversarial attacks for deep neural networks produced a large variety of algorithms for training robust models. Most of the effective algorithms involve solving the min-max optimization problem for training robust models (min step) under worst-case attacks (max step). However, they often suffer from high computational cost from running several inner maximization iterations (to find an optimal attack) inside every outer minimization iteration. Therefore, it becomes difficult to readily apply such algorithms for moderate to large size real world data sets. To alleviate this, we explore the effectiveness of iterative descent-ascent algorithms where the maximization and minimization steps are executed in an alternate fashion to simultaneously obtain the worst-case attack and the corresponding robust model. Specifically, we propose a novel discrete-time dynamical system-based algorithm that aims to find the saddle point of a min-max optimization problem in the presence of uncertainties. Under the assumptions that the cost function is convex and uncertainties enter concavely in the robust learning problem, we analytically show that our algorithm converges asymptotically to the robust optimal solution under a general adversarial budget constraints as induced by ℓp norm, for 1≤p≤∞. Based on our proposed analysis, we devise a fast robust training algorithm for deep neural networks. Although such training involves highly non-convex robust optimization problems, empirical results show that the algorithm can achieve significant robustness compared to other state-of-the-art robust models on benchmark data sets.

Fast inverse design of microstructures via generative invariance networks.

The problem of the efficient design of material microstructures exhibiting desired properties spans a variety of engineering and science applications. The ability to rapidly generate microstructures that exhibit user-specified property distributions can transform the iterative process of traditional microstructure-sensitive design. We reformulate the microstructure design process using a constrained generative adversarial network (GAN) model. This approach explicitly encodes invariance constraints within GANs to generate two-phase morphologies for photovoltaic applications obeying design specifications: specifically, user-defined short-circuit current density and fill factor combinations. Such invariance constraints can be represented by differentiable, deep learning-based surrogates of full physics models mapping microstructures to photovoltaic properties. Furthermore, we propose a multi-fidelity surrogate that reduces expensive label requirements by a factor of five. Our framework enables the incorporation of expensive or non-differentiable constraints for the fast generation of microstructures (in 190 ms) with user-defined properties. Such proposed physics-aware data-driven methods for inverse design problems can be used to considerably accelerate the field of microstructure-sensitive design.

A Deep Learning Framework for Design and Analysis of Surgical Bioprosthetic Heart Valves.

Bioprosthetic heart valves (BHVs) are commonly used as heart valve replacements but they are prone to fatigue failure; estimating their remaining life directly from medical images is difficult. Analyzing the valve performance can provide better guidance for personalized valve design. However, such analyses are often computationally intensive. In this work, we introduce the concept of deep learning (DL) based finite element analysis (DLFEA) to learn the deformation biomechanics of bioprosthetic aortic valves directly from simulations. The proposed DL framework can eliminate the time-consuming biomechanics simulations, while predicting valve deformations with the same fidelity. We present statistical results that demonstrate the high performance of the DLFEA framework and the applicability of the framework to predict bioprosthetic aortic valve deformations. With further development, such a tool can provide fast decision support for designing surgical bioprosthetic aortic valves. Ultimately, this framework could be extended to other BHVs and improve patient care.