Project description:The geometric designs of MEMS devices can profoundly impact their physical properties and eventual performances. However, it is challenging for researchers to rationally consider a large number of possible designs, as it would be very time- and resource-consuming to study all these cases using numerical simulation. In this paper, we report the use of deep learning techniques to accelerate the MEMS design cycle by quickly and accurately predicting the physical properties of numerous design candidates with vastly different geometric features. Design candidates are represented in a nonparameterized, topologically unconstrained form using pixelated black-and-white images. After sufficient training, a deep neural network can quickly calculate the physical properties of interest with good accuracy without using conventional numerical tools such as finite element analysis. As an example, we apply our deep learning approach in the prediction of the modal frequency and quality factor of disk-shaped microscale resonators. With reasonable training, our deep learning neural network becomes a high-speed, high-accuracy calculator: it can identify the flexural mode frequency and the quality factor 4.6 × 103 times and 2.6 × 104 times faster, respectively, than conventional numerical simulation packages, with good accuracies of 98.8 ± 1.6% and 96.8 ± 3.1%, respectively. When simultaneously predicting the frequency and the quality factor, up to ~96.0% of the total computation time can be saved during the design process. The proposed technique can rapidly screen over thousands of design candidates and promotes experience-free and data-driven MEMS structural designs.
Project description:In this paper, we discuss parameterized algorithms for variants of the partial vertex cover problem. Recall that in the classical vertex cover problem (VC), we are given a graph
Project description:In this paper, we leverage over-parameterization to design regularization-free algorithms for the high-dimensional single index model and provide theoretical guarantees for the induced implicit regularization phenomenon. Specifically, we study both vector and matrix single index models where the link function is nonlinear and unknown, the signal parameter is either a sparse vector or a low-rank symmetric matrix, and the response variable can be heavy-tailed. To gain a better understanding of the role played by implicit regularization without excess technicality, we assume that the distribution of the covariates is known a priori. For both the vector and matrix settings, we construct an over-parameterized least-squares loss function by employing the score function transform and a robust truncation step designed specifically for heavy-tailed data. We propose to estimate the true parameter by applying regularization-free gradient descent to the loss function. When the initialization is close to the origin and the stepsize is sufficiently small, we prove that the obtained solution achieves minimax optimal statistical rates of convergence in both the vector and matrix cases. In addition, our experimental results support our theoretical findings and also demonstrate that our methods empirically outperform classical methods with explicit regularization in terms of both ℓ2-statistical rate and variable selection consistency.
Project description:We consider parameterized concurrent systems consisting of a finite but unknown number of components, obtained by replicating a given set of finite state automata.
Project description:Learning from demonstration (LfD) enables a robot to emulate natural human movement instead of merely executing preprogrammed behaviors. This article presents a hierarchical LfD structure of task-parameterized models for object movement tasks, which are ubiquitous in everyday life and could benefit from robotic support. Our approach uses the task-parameterized Gaussian mixture model (TP-GMM) algorithm to encode sets of demonstrations in separate models that each correspond to a different task situation. The robot then maximizes its expected performance in a new situation by either selecting a good existing model or requesting new demonstrations. Compared to a standard implementation that encodes all demonstrations together for all test situations, the proposed approach offers four advantages. First, a simply defined distance function can be used to estimate test performance by calculating the similarity between a test situation and the existing models. Second, the proposed approach can improve generalization, e.g., better satisfying the demonstrated task constraints and speeding up task execution. Third, because the hierarchical structure encodes each demonstrated situation individually, a wider range of task situations can be modeled in the same framework without deteriorating performance. Last, adding or removing demonstrations incurs low computational load, and thus, the robot's skill library can be built incrementally. We first instantiate the proposed approach in a simulated task to validate these advantages. We then show that the advantages transfer to real hardware for a task where naive participants collaborated with a Willow Garage PR2 robot to move a handheld object. For most tested scenarios, our hierarchical method achieved significantly better task performance and subjective ratings than both a passive model with only gravity compensation and a single TP-GMM encoding all demonstrations.
Project description:Peripheral nerve stimulation is an effective treatment for various neurological disorders. The method of activation and stimulation parameters used impact the efficacy of the therapy, which emphasizes the need for tools to model this behavior. Computational modeling of nerve stimulation has proven to be a useful tool for estimating stimulation thresholds, optimizing electrode design, and exploring previously untested stimulation methods. Despite their utility, these tools require access to and familiarity with several pieces of specialized software. A simpler, streamlined process would increase accessibility significantly. We developed an open-source, parameterized model with a simple online user interface that allows user to adjust up to 36 different parameters (https://nervestimlab.utdallas.edu). The model accurately predicts fiber activation thresholds for nerve and electrode combinations reported in literature. Additionally, it replicates characteristic differences between stimulation methods, such as lower thresholds with monopolar stimulation as compared to tripolar stimulation. The model predicted that the difference in threshold between monophasic and biphasic waveforms, a well-characterized phenomenon, is not present during stimulation with bipolar electrodes.In vivotesting on the rat sciatic nerve validated this prediction, which has not been previously reported. The accuracy of the model when compared to previous experiments, as well as the ease of use and accessibility to generate testable hypotheses, indicate that this software may represent a useful tool for a variety of nerve stimulation applications.
Project description:The mutation-selection model of coding sequence evolution has received renewed attention for its use in estimating site-specific amino acid propensities and selection coefficient distributions. Two computationally tractable mutation-selection inference frameworks have been introduced: One framework employs a fixed-effects, highly parameterized maximum likelihood approach, whereas the other employs a random-effects Bayesian Dirichlet Process approach. While both implementations follow the same model, they appear to make distinct predictions about the distribution of selection coefficients. The fixed-effects framework estimates a large proportion of highly deleterious substitutions, whereas the random-effects framework estimates that all substitutions are either nearly neutral or weakly deleterious. It remains unknown, however, how accurately each method infers evolutionary constraints at individual sites. Indeed, selection coefficient distributions pool all site-specific inferences, thereby obscuring a precise assessment of site-specific estimates. Therefore, in this study, we use a simulation-based strategy to determine how accurately each approach recapitulates the selective constraint at individual sites. We find that the fixed-effects approach, despite its extensive parameterization, consistently and accurately estimates site-specific evolutionary constraint. By contrast, the random-effects Bayesian approach systematically underestimates the strength of natural selection, particularly for slowly evolving sites. We also find that, despite the strong differences between their inferred selection coefficient distributions, the fixed- and random-effects approaches yield surprisingly similar inferences of site-specific selective constraint. We conclude that the fixed-effects mutation-selection framework provides the more reliable software platform for model application and future development.
Project description:Many applications in protein engineering require optimizing multiple protein properties simultaneously, such as binding one target but not others or binding a target while maintaining stability. Such multistate design problems require navigating a high-dimensional space to find proteins with desired characteristics. A model that relates protein sequence to functional attributes can guide design to solutions that would be hard to discover via screening. In this work, we measured thousands of protein-peptide binding affinities with the high-throughput interaction assay amped SORTCERY and used the data to parameterize a model of the alpha-helical peptide-binding landscape for three members of the Bcl-2 family of proteins: Bcl-xL, Mcl-1, and Bfl-1. We applied optimization protocols to explore extremes in this landscape to discover peptides with desired interaction profiles. Computational design generated 36 peptides, all of which bound with high affinity and specificity to just one of Bcl-xL, Mcl-1, or Bfl-1, as intended. We designed additional peptides that bound selectively to two out of three of these proteins. The designed peptides were dissimilar to known Bcl-2-binding peptides, and high-resolution crystal structures confirmed that they engaged their targets as expected. Excellent results on this challenging problem demonstrate the power of a landscape modeling approach, and the designed peptides have potential uses as diagnostic tools or cancer therapeutics.
Project description:We applied the high-throughput interaction assay SORTCERY to measure thousands of protein-peptide binding affinities and used the data to parameterize models of the peptide-binding landscape for three members of the Bcl-2 family of proteins. We applied the models to design peptides that bound with high affinity and specificity to just one of Bcl-xL, Mcl-1, or Bfl-1. We designed additional peptides that bound selectively to two out of three of these proteins. The raw data provided are the multiplexed fastq files that serve as inputs to our analysis pipeline. Additional detail is available in our corresponding publication and at the following github repository. https://github.com/KeatingLab/sortcery_design
Project description:Studies of economic mobility summarize the distribution of offspring incomes for each level of parent income. Mitnik and Grusky (2020) highlight that the conventional intergenerational elasticity (IGE) targets the geometric mean and propose a parametric strategy for estimating the arithmetic mean. We decompose the IGE and their proposal into two choices: (1) the summary statistic for the conditional distribution and (2) the functional form. These choices lead us to a different strategy-visualizing several quantiles of the offspring income distribution as smooth functions of parent income. Our proposal solves the problems Mitnik and Grusky highlight with geometric means, avoids the sensitivity of arithmetic means to top incomes, and provides more information than is possible with any single number. Our proposal has broader implications: the default summary (the mean) used in many regressions is sensitive to the tail of the distribution in ways that may be substantively undesirable.