Learning Complexity to Guide Light-Induced Self-Organized Nanopatterns.

Ultrafast laser irradiation can induce spontaneous self-organization of surfaces into dissipative structures with nanoscale reliefs. These surface patterns emerge from symmetry-breaking dynamical processes that occur in Rayleigh-Bénard-like instabilities. In this study, we demonstrate that the coexistence and competition between surface patterns of different symmetries in two dimensions can be numerically unraveled using the stochastic generalized Swift-Hohenberg model. We originally propose a deep convolutional network to identify and learn the dominant modes that stabilize for a given bifurcation and quadratic model coefficients. The model is scale-invariant and has been calibrated on microscopy measurements using a physics-guided machine learning strategy. Our approach enables the identification of experimental irradiation conditions for a desired self-organization pattern. It can be generally applied to predict structure formation in situations where the underlying physics can be approximately described by a self-organization process and data is sparse and nontime series. Our Letter paves the way for supervised local manipulation of matter using timely controlled optical fields in laser manufacturing.

Learning PDE to Model Self-Organization of Matter.

A self-organization hydrodynamic process has recently been proposed to partially explain the formation of femtosecond laser-induced nanopatterns on Nickel, which have important applications in optics, microbiology, medicine, etc. Exploring laser pattern space is difficult, however, which simultaneously (i) motivates using machine learning (ML) to search for novel patterns and (ii) hinders it, because of the few data available from costly and time-consuming experiments. In this paper, we use ML to predict novel patterns by integrating partial physical knowledge in the form of the Swift-Hohenberg (SH) partial differential equation (PDE). To do so, we propose a framework to learn with few data, in the absence of initial conditions, by benefiting from background knowledge in the form of a PDE solver. We show that in the case of a self-organization process, a feature mapping exists in which initial conditions can safely be ignored and patterns can be described in terms of PDE parameters alone, which drastically simplifies the problem. In order to apply this framework, we develop a second-order pseudospectral solver of the SH equation which offers a good compromise between accuracy and speed. Our method allows us to predict new nanopatterns in good agreement with experimental data. Moreover, we show that pattern features are related, which imposes constraints on novel pattern design, and suggest an efficient procedure of acquiring experimental data iteratively to improve the generalization of the learned model. It also allows us to identify the limitations of the SH equation as a partial model and suggests an improvement to the physical model itself.

Learning personalized ADL recognition models from few raw data.

Recognition of activities of daily living (ADL) is an essential component of assisted living systems based on actigraphy. This task can nowadays be performed by machine learning models which are able to automatically extract and learn relevant features but, most of time, need to be trained with large amounts of data collected on several users. In this paper, we propose an approach to learn personalized ADL recognition models from few raw data based on a specific type of neural network called matching network. The interest of this few-shot learning approach is three-fold. Firstly, people perform activities their own way and general models may average out important individual characteristics unlike personalized models that could thus achieve better performance. Secondly, gathering large quantities of annotated data from one user is time-consuming and threatens privacy in a medical context. Thirdly, matching networks are by nature weakly dependent on the classes they are trained on and can generalize easily to new activities without needing extra training, thus making them very versatile for real applications. Our results show the effectiveness of the proposed approach compared to general neural network models, even in situations with few training data.

Low-Complexity Approximate Convolutional Neural Networks.

In this paper, we present an approach for minimizing the computational complexity of the trained convolutional neural networks (ConvNets). The idea is to approximate all elements of a given ConvNet and replace the original convolutional filters and parameters (pooling and bias coefficients; and activation function) with an efficient approximations capable of extreme reductions in computational complexity. Low-complexity convolution filters are obtained through a binary (zero and one) linear programming scheme based on the Frobenius norm over sets of dyadic rationals. The resulting matrices allow for multiplication-free computations requiring only addition and bit-shifting operations. Such low-complexity structures pave the way for low power, efficient hardware designs. We applied our approach on three use cases of different complexities: 1) a "light" but efficient ConvNet for face detection (with around 1000 parameters); 2) another one for hand-written digit classification (with more than 180 000 parameters); and 3) a significantly larger ConvNet: AlexNet with million matrices. We evaluated the overall performance on the respective tasks for different levels of approximations. In all considered applications, very low-complexity approximations have been derived maintaining an almost equal classification performance.

Pairwise Identity Verification via Linear Concentrative Metric Learning.

This paper presents a study of metric learning systems on pairwise identity verification, including pairwise face verification and pairwise speaker verification, respectively. These problems are challenging because the individuals in training and testing are mutually exclusive, and also due to the probable setting of limited training data. For such pairwise verification problems, we present a general framework of metric learning systems and employ the stochastic gradient descent algorithm as the optimization solution. We have studied both similarity metric learning and distance metric learning systems, of either a linear or shallow nonlinear model under both restricted and unrestricted training settings. Extensive experiments demonstrate that with limited training pairs, learning a linear system on similar pairs only is preferable due to its simplicity and superiority, i.e., it generally achieves competitive performance on both the labeled faces in the wild face dataset and the NIST speaker dataset. It is also found that a pretrained deep nonlinear model helps to improve the face verification results significantly.

Fast Pixelwise Adaptive Visual Tracking of Non-Rigid Objects.

In this paper, we present a new algorithm for real-time single-object tracking in videos in unconstrained environments. The algorithm comprises two different components that are trained "in one shot" at the first video frame: a detector that makes use of the generalized Hough transform with color and gradient descriptors and a probabilistic segmentation method based on global models for foreground and background color distributions. Both components work at pixel level and are used for tracking in a combined way adapting each other in a co-training manner. Moreover, we propose an adaptive shape model as well as a new probabilistic method for updating the scale of the tracker. Through effective model adaptation and segmentation, the algorithm is able to track objects that undergo rigid and non-rigid deformations and considerable shape and appearance variations. The proposed tracking method has been thoroughly evaluated on challenging benchmarks, and outperforms the state-of-the-art tracking methods designed for the same task. Finally, a very efficient implementation of the proposed models allows for extremely fast tracking.

Track creation and deletion framework for long-term online multiface tracking.

To improve visual tracking, a large number of papers study more powerful features, or better cue fusion mechanisms, such as adaptation or contextual models. A complementary approach consists of improving the track management, that is, deciding when to add a target or stop its tracking, for example, in case of failure. This is an essential component for effective multiobject tracking applications, and is often not trivial. Deciding whether or not to stop a track is a compromise between avoiding erroneous early stopping while tracking is fine, and erroneous continuation of tracking when there is an actual failure. This decision process, very rarely addressed in the literature, is difficult due to object detector deficiencies or observation models that are insufficient to describe the full variability of tracked objects and deliver reliable likelihood (tracking) information. This paper addresses the track management issue and presents a real-time online multiface tracking algorithm that effectively deals with the above difficulties. The tracking itself is formulated in a multiobject state-space Bayesian filtering framework solved with Markov Chain Monte Carlo. Within this framework, an explicit probabilistic filtering step decides when to add or remove a target from the tracker, where decisions rely on multiple cues such as face detections, likelihood measures, long-term observations, and track state characteristics. The method has been applied to three challenging data sets of more than 9 h in total, and demonstrate a significant performance increase compared to more traditional approaches (Markov Chain Monte Carlo, reversible-jump Markov Chain Monte Carlo) only relying on head detection and likelihood for track management.