skip to main content

Title: Dense Crowd Counting Convolutional Neural Networks with Minimal Data using Semi-Supervised Dual-Goal Generative Adversarial Networks
In this work, we generalize semi-supervised generative adversarial networks (GANs) from classification problems to regression for use in dense crowd counting. In the last several years, the importance of improving the training of neural networks using semi-supervised training has been thoroughly demonstrated for classification problems. This work presents a dual-goal GAN which seeks both to provide the number of individuals in a densely crowded scene and distinguish between real and generated images. This method allows the dual-goal GAN to benefit from unlabeled data in the training process, improving the predictive capabilities of the discriminating network compared to the fully-supervised version of the network. Typical semi-supervised GANs are unable to function in the regression regime due to biases introduced when using a single prediction goal. Using the proposed approach, the amount of data which needs to be annotated for dense crowd counting can be significantly reduced.
Authors:
; ; ;
Award ID(s):
1827505 1737533
Publication Date:
NSF-PAR ID:
10110611
Journal Name:
IEEE Conference on Computer Vision and Pattern Recognition: Learning with Imperfect Data Workshop
Page Range or eLocation-ID:
21-28
Sponsoring Org:
National Science Foundation
More Like this
  1. In this work, we use a generative adversarial network (GAN) to train crowd counting networks using minimal data. We describe how GAN objectives can be modified to allow for the use of unlabeled data to benefit inference training in semi-supervised learning. More generally, we explain how these same methods can be used in more generic multiple regression target semi-supervised learning, with crowd counting being a demonstrative example. Given a convolutional neural network (CNN) with capabilities equivalent to the discriminator in the GAN, we provide experimental results which show that our GAN is able to outperform the CNN even when the CNN has access to significantly more labeled data. This presents the potential of training such networks to high accuracy with little data. Our primary goal is not to outperform the state-of-the-art using an improved method on the entire dataset, but instead we work to show that through semi-supervised learning we can reduce the data required to train an inference network to a given accuracy. To this end, systematic experiments are performed with various numbers of images and cameras to show under which situations the semi-supervised GANs can improve results.
  2. A broad class of unsupervised deep learning methods such as Generative Adversarial Networks (GANs) involve training of overparameterized models where the number of parameters of the model exceeds a certain threshold. Indeed, most successful GANs used in practice are trained using overparameterized generator and discriminator networks, both in terms of depth and width. A large body of work in supervised learning have shown the importance of model overparameterization in the convergence of the gradient descent (GD) to globally optimal solutions. In contrast, the unsupervised setting and GANs in particular involve non-convex concave mini-max optimization problems that are often trained using Gradient Descent/Ascent (GDA). The role and benefits of model overparameterization in the convergence of GDA to a global saddle point in non-convex concave problems is far less understood. In this work, we present a comprehensive analysis of the importance of model overparameterization in GANs both theoretically and empirically. We theoretically show that in an overparameterized GAN model with a 1-layer neural network generator and a linear discriminator, GDA converges to a global saddle point of the underlying non-convex concave min-max problem. To the best of our knowledge, this is the first result for global convergence of GDA in such settings.more »Our theory is based on a more general result that holds for a broader class of nonlinear generators and discriminators that obey certain assumptions (including deeper generators and random feature discriminators). Our theory utilizes and builds upon a novel connection with the convergence analysis of linear timevarying dynamical systems which may have broader implications for understanding the convergence behavior of GDA for non-convex concave problems involving overparameterized models. We also empirically study the role of model overparameterization in GANs using several large-scale experiments on CIFAR-10 and Celeb-A datasets. Our experiments show that overparameterization improves the quality of generated samples across various model architectures and datasets. Remarkably, we observe that overparameterization leads to faster and more stable convergence behavior of GDA across the board.« less
  3. Deep neural networks have become increasingly popular in radar micro-Doppler classification; yet, a key challenge, which has limited potential gains, is the lack of large amounts of measured data that can facilitate the design of deeper networks with greater robustness and performance. Several approaches have been proposed in the literature to address this problem, such as unsupervised pre-training and transfer learning from optical imagery or synthetic RF data. This work investigates an alternative approach to training which involves exploitation of “datasets of opportunity” – micro-Doppler datasets collected using other RF sensors, which may be of a different frequency, bandwidth or waveform - for the purposes of training. Specifically, this work compares in detail the cross-frequency training degradation incurred for several different training approaches and deep neural network (DNN) architectures. Results show a 70% drop in classification accuracy when the RF sensors for pre-training, fine-tuning, and testing are different, and a 15% degradation when only the pre-training data is different, but the fine-tuning and test data are from the same sensor. By using generative adversarial networks (GANs), a large amount of synthetic data is generated for pre-training. Results show that cross-frequency performance degradation is reduced by 50% when kinematically-sifted GAN-synthesized signaturesmore »are used in pre-training.« less
  4. Generative Adversarial Networks (GANs) recently demonstrated a great opportunity toward unsupervised learning with the intention to mitigate the massive human efforts on data labeling in supervised learning algorithms. GAN combines a generative model and a discriminative model to oppose each other in an adversarial situation to refine their abilities. Existing nonvolatile memory based machine learning accelerators, however, could not support the computational needs required by GAN training. Specifically, the generator utilizes a new operator, called transposed convolution, which introduces significant resource underutilization when executed on conventional neural network accelerators as it inserts massive zeros in its input before a convolution operation. In this work, we propose a novel computational deformation technique that synergistically optimizes the forward and backward functions in transposed convolution to eliminate the large resource underutilization. In addition, we present dedicated control units - a dataflow mapper and an operation scheduler, to support the proposed execution model with high parallelism and low energy consumption. ZARA is implemented with commodity ReRAM chips, and experimental results show that our design can improve GAN’s training performance by averagely 1.6x~23x over CMOS-based GAN accelerators. Compared to state-of-the-art ReRAM-based accelerator designs, ZARA also provides 1.15x~2.1x performance improvement.
  5. We describe the convex semi-infinite dual of the two-layer vector-output ReLU neural network training problem. This semi-infinite dual admits a finite dimensional representation, but its support is over a convex set which is difficult to characterize. In particular, we demonstrate that the non-convex neural network training problem is equivalent to a finite-dimensional convex copositive program. Our work is the first to identify this strong connection between the global optima of neural networks and those of copositive programs. We thus demonstrate how neural networks implicitly attempt to solve copositive programs via semi-nonnegative matrix factorization, and draw key insights from this formulation. We describe the first algorithms for provably finding the global minimum of the vector output neural network training problem, which are polynomial in the number of samples for a fixed data rank, yet exponential in the dimension. However, in the case of convolutional architectures, the computational complexity is exponential in only the filter size and polynomial in all other parameters. We describe the circumstances in which we can find the global optimum of this neural network training problem exactly with soft-thresholded SVD, and provide a copositive relaxation which is guaranteed to be exact for certain classes of problems, and whichmore »corresponds with the solution of Stochastic Gradient Descent in practice.« less