Hierarchical neural networks are exponentially more efficient than their corresponding "shallow" counterpart with the same expressive power, but involve huge number of parameters and require tedious amounts of training. By approximating the tangent subspace, we suggest a sparse representation that enables switching to shallow networks, GradNet after a very early training stage. Our experiments show that the proposed approximation of the metric improves and sometimes even surpasses the achievable performance of the original network significantly even after a few epochs of training the original feedforward network.

An Intrusion Detection System (IDS) is a key cybersecurity tool for network administrators as it identifies malicious traffic and cyberattacks. With the recent successes of machine learning techniques such as deep learning, more and more IDS are now using machine learning algorithms to detect attacks faster. However, these systems lack robustness when facing previously unseen types of attacks. With the increasing number of new attacks, especially against Internet of Things devices, having a robust IDS able to spot unusual and new attacks becomes necessary. This work explores the possibility of leveraging generative adversarial models to improve the robustness of machine learning based IDS. More specifically, we propose a new method named SIGMA, that leverages adversarial examples to strengthen IDS against new types of attacks. Using Generative Adversarial Networks (GAN) and metaheuristics, SIGMA %Our method consists in generates adversarial examples, iteratively, and uses it to retrain a machine learning-based IDS, until a convergence of the detection rate (i.e. until the detection system is not improving anymore). A round of improvement consists of a generative phase, in which we use GANs and metaheuristics to generate instances ; an evaluation phase in which we calculate the detection rate of those newly generated attacks ; and a training phase, in which we train the IDS with those attacks. We have evaluated the SIGMA method for four standard machine learning classification algorithms acting as IDS, with a combination of GAN and a hybrid local-search and genetic algorithm, to generate new datasets of attacks. Our results show that SIGMA can successfully generate adversarial attacks against different machine learning based IDS. Also, using SIGMA, we can improve the performance of an IDS to up to 100\% after as little as two rounds of improvement.

This paper explores the scenarios under which an attacker can claim that 'Noise and access to the softmax layer of the model is all you need' to steal the weights of a convolutional neural network whose architecture is already known. We were able to achieve 96% test accuracy using the stolen MNIST model and 82% accuracy using the stolen KMNIST model learned using only i.i.d. Bernoulli noise inputs. We posit that this theft-susceptibility of the weights is indicative of the complexity of the dataset and propose a new metric that captures the same. The goal of this dissemination is to not just showcase how far knowing the architecture can take you in terms of model stealing, but to also draw attention to this rather idiosyncratic weight learnability aspects of CNNs spurred by i.i.d. noise input. We also disseminate some initial results obtained with using the Ising probability distribution in lieu of the i.i.d. Bernoulli distribution.

The interplay between inter-neuronal network topology and cognition has been studied deeply by connectomics researchers and network scientists, which is crucial towards understanding the remarkable efficacy of biological neural networks. Curiously, the deep learning revolution that revived neural networks has not paid much attention to topological aspects. The architectures of deep neural networks (DNNs) do not resemble their biological counterparts in the topological sense. We bridge this gap by presenting initial results of Deep Connectomics Networks (DCNs) as DNNs with topologies inspired by real-world neuronal networks. We show high classification accuracy obtained by DCNs whose architecture was inspired by the biological neuronal networks of C. Elegans and the mouse visual cortex.

This paper investigates multilevel initialization strategies for training very deep neural networks with a layer-parallel multigrid solver. The scheme is based on the continuous interpretation of the training problem as a problem of optimal control, in which neural networks are represented as discretizations of time-dependent ordinary differential equations. A key goal is to develop a method able to intelligently initialize the network parameters for the very deep networks enabled by scalable layer-parallel training. To do this, we apply a refinement strategy across the time domain, that is equivalent to refining in the layer dimension. The resulting refinements create deep networks, with good initializations for the network parameters coming from the coarser trained networks. We investigate the effectiveness of such multilevel "nested iteration" strategies for network training, showing supporting numerical evidence of reduced run time for equivalent accuracy. In addition, we study whether the initialization strategies provide a regularizing effect on the overall training process and reduce sensitivity to hyperparameters and randomness in initial network parameters.

Machine Learning (ML) is widely used for several tasks with time-series and biosensor data such as for human activity recognition, electronic health records data-based predictions (Ismail Fawaz et al., 2019), and real-time bionsensor-based decisions. V arious classification goals are addressed related to electrocardiography (ECG) (Jambukia et al., 2015), elec-troencephalography (EEG) (Craik et al., 2019; Dose et al., 2018), and electromyograpy (EMG) (Ketyk et al., 2019; Hu et al., 2018; Patricia et al., 2014; Du et al., 2017). Sensing hand gestures can be done by means of wearables or by means of image or video analysis of hand or finger motion. A wearable-based detection can physically rely on measuring the acceleration and rotations of our body parts (arms, hands or fingers) with Inertial Measurement Unit (IMU) sensors or by measuring the myo-electric signals generated by the various muscles of our arms or fingers with EMG sensors. Surface EMG (sEMG) records muscle activity from the surface of the skin which is above the muscle being evaluated. The signal is collected via surface electrodes. We are interested in sEMG-sensor placement to the forearm and performing hand gesture recognition with ML.

The success of convolutional neural networks (CNNs) in various applications is accompanied by a significant increase in computation and parameter storage costs. Recent efforts to reduce these overheads involve pruning and compressing the weights of various layers while at the same time aiming to not sacrifice performance. In this paper, we propose a novel criterion for CNN pruning inspired by neural network interpretability: The most relevant elements, i.e. weights or filters, are automatically found using their relevance score in the sense of explainable AI (XAI). By that we for the first time link the two disconnected lines of interpretability and model compression research. We show in particular that our proposed method can efficiently prune transfer-learned CNN models where networks pre-trained on large corpora are adapted to specialized tasks. To this end, the method is evaluated on a broad range of computer vision datasets. Notably, our novel criterion is not only competitive or better compared to state-of-the-art pruning criteria when successive retraining is performed, but clearly outperforms these previous criteria in the common application setting where the data of the task to be transferred to are very scarce and no retraining is possible. Our method can iteratively compress the model while maintaining or even improving accuracy. At the same time, it has a computational cost in the order of gradient computation and is comparatively simple to apply without the need for tuning hyperparameters for pruning.

This work presents an analysis of the discriminators used in Generative Adversarial Networks (GANs) for Video. We show that unconstrained video discriminator architectures induce a loss surface with high curvature which make optimisation difficult. We also show that this curvature becomes more extreme as the maximal kernel dimension of video discriminators increases. With these observations in hand, we propose a family of efficient Lower-Dimensional Video Discriminators for GANs (LDVD GANs). The proposed family of discriminators improve the performance of video GAN models they are applied to and demonstrate good performance on complex and diverse datasets such as UCF-101. In particular, we show that they can double the performance of Temporal-GANs and provide for state-of-the-art performance on a single GPU.

We introduce DeepInversion, a new method for synthesizing images from the image distribution used to train a deep neural network. We 'invert' a trained network (teacher) to synthesize class-conditional input images starting from random noise, without using any additional information about the training dataset. Keeping the teacher fixed, our method optimizes the input while regularizing the distribution of intermediate feature maps using information stored in the batch normalization layers of the teacher. Further, we improve the diversity of synthesized images using Adaptive DeepInversion, which maximizes the Jensen-Shannon divergence between the teacher and student network logits. The resulting synthesized images from networks trained on the CIFAR-10 and ImageNet datasets demonstrate high fidelity and degree of realism, and help enable a new breed of data-free applications - ones that do not require any real images or labeled data. We demonstrate the applicability of our proposed method to three tasks of immense practical importance -- (i) data-free network pruning, (ii) data-free knowledge transfer, and (iii) data-free continual learning.

Convolutional neural networks were recently employed to fully reconstruct fluid simulation data from a set of reduced parameters. However, since (de-)convolutions traditionally trained with supervised L1-loss functions do not discriminate between low and high frequencies in the data, the error is not minimized efficiently for higher bands. This directly correlates with the quality of the perceived results, since missing high frequency details are easily noticeable. In this paper, we analyze the reconstruction quality of generative networks and present a frequency-aware loss function that is able to focus on specific bands of the dataset during training time. We show that our approach improves reconstruction quality of fluid simulation data in mid-frequency bands, yielding perceptually better results while requiring comparable training time.