In this project, we are going to use a pre-trained VGG16 model which looks as follows. Keep in mind that we are not going to use fully connected (blue) and softmax layers (yellow). They act as a classifier which we don't need here. We are going to use only feature extractors i.e convolutional (black) and max pooling (red) layers. Let's take a look at how specific features look at the selected VGG16 layers trained on ImageNet dataset. We are not going to visualize every CNN layer here but according to Johnson et al. for a content layer we should select While this combination is proved to be working I recommend you to play with it and experiment with different layers.

In most of the practical use cases, data scientists are satisfied by machine learning models that simply make predictions. Given unseen observations, a model performs the prediction of certain outcome . The performance of such a model is usually assessed by comparing the predicted value with the ground truth, whenever available. There is however, another measure that might be of interest: uncertainty. How uncertain is a model in predicting a particular sample?

The purpose of this project is to introduce a shortcut to developers and researcher for finding useful resources about Deep Learning. There are different motivations for this open source project. There other similar repositories similar to this repository and are very comprehensive and useful and to be honest they made me ponder if there is a necessity for this repository! The point of this repository is that the resources are being targeted. The organization of the resources is such that the user can easily find the things he/she is looking for.

Artificial intelligence initiatives are springing up in almost every industry and generating a huge market in their wake. Gartner predicts that AI augmentation will generate $3.9 trillion in business value by 2022 alone. What's more, Gartner says that AI promises to be the most disruptive class of technologies during the next 10 years, driven by increases in computational power, advances in storage technology, the availability of new data and the ubiquity of deep learning toolkits. Organizations making the journey to AI face a multitude of complex choices related to data, skillsets, software stacks, analytic toolkits and infrastructure components. Each of these choices has significant implications for the time to value associated with AI initiatives.

The long, arduous process of narrowing down millions of chemical compounds to just a select few that can be further developed into mature drugs, may soon be shortened, thanks to new artificial intelligence (AI) software. Genedata, a bioinformatics solutions company headquartered in Basel, Switzerland, has created Imagence, a high content screening image analysis workflow based on deep learning that cuts image analysis times, while increasing data quality and reproducibility of results. "We have software systems which can more or less analyze almost every assay that you need there, can construct and organize the data, store the data, federate the data and make a decisions along this process," said Stephen Steigele, the head of science at Genedata said in an exclusive interview with R&D Magazine "What we have now specifically solved is we developed a software where we use artificial intelligence to make a part of this research process extremely easy." The task of analyzing high content screening images is often labor-intensive and time-consuming, involving several different levels of expertise with several manual steps, like the selection of extracted features or correct detection of cells. This process, which can take many weeks, is reduced to only a few hours using the new technology.

This course is an introduction to Neural Networks, so you need absolutely no prior knowledge in Artificial Intelligence, Machine Learning, and AI. However, you need to have basic understanding of programming specially in Java to easily follow the coding video. If you just want to lean the mathematical model and the problem solving process using Neural Networks, you can then skip the coding videos. Machine learning is an extremely hot area in Artificial Intelligence and Data Science. There is no doubt that Neural Networks are the most well-regarded and widely used machine learning techniques. A lot of Data Scientists use Neural Networks without understanding their internal structure.

Previous work on predicting or generating 3D human pose sequences regresses either joint rotations or joint positions. The former strategy is prone to error accumulation along the kinematic chain, as well as discontinuities when using Euler angles or exponential maps as parameterizations. The latter requires re-projection onto skeleton constraints to avoid bone stretching and invalid configurations. This work addresses both limitations. QuaterNet represents rotations with quaternions and our loss function performs forward kinematics on a skeleton to penalize absolute position errors instead of angle errors. We investigate both recurrent and convolutional architectures and evaluate on short-term prediction and long-term generation. For the latter, our approach is qualitatively judged as realistic as recent neural strategies from the graphics literature. Our experiments compare quaternions to Euler angles as well as exponential maps and show that only a very short context is required to make reliable future predictions. Finally, we show that the standard evaluation protocol for Human3.6M produces high variance results and we propose a simple solution.

This paper presents an unsupervised method to learn a neural network, namely an explainer, to interpret a pre-trained convolutional neural network (CNN), i.e., the explainer uses interpretable visual concepts to explain features in middle conv-layers of a CNN. Given feature maps of a conv-layer of the CNN, the explainer performs like an auto-encoder, which decomposes the feature maps into object-part features. The object-part features are learned to reconstruct CNN features without much loss of information. We can consider the disentangled representations of object parts a paraphrase of CNN features, which help people understand the knowledge encoded by the CNN. More crucially, we learn the explainer via knowledge distillation without using any annotations of object parts or textures for supervision. In experiments, our method was widely used to interpret features of different benchmark CNNs, and explainers significantly boosted the feature interpretability without hurting the discrimination power of the CNNs.

Deep neural networks (DNNs) have achieved superior performance in various prediction tasks, but can be very vulnerable to adversarial examples or perturbations. Therefore, it is crucial to measure the sensitivity of DNNs to various forms of perturbations in real applications. We introduce a novel perturbation manifold and its associated influence measure to quantify the effects of various perturbations on DNN classifiers. Such perturbations include various external and internal perturbations to input samples and network parameters. The proposed measure is motivated by information geometry and provides desirable invariance properties. We demonstrate that our influence measure is useful for four model building tasks: detecting potential 'outliers', analyzing the sensitivity of model architectures, comparing network sensitivity between training and test sets, and locating vulnerable areas. Experiments show reasonably good performance of the proposed measure for the popular DNN models ResNet50 and DenseNet121 on CIFAR10 and MNIST datasets.

Recently, deep learning based natural language processing techniques are being extensively used to deal with spam mail, censorship evaluation in social networks, among others. However, there is only a couple of works evaluating the vulnerabilities of such deep neural networks. Here, we go beyond attacks to investigate, for the first time, universal rules, i.e., rules that are sample agnostic and therefore could turn any text sample in an adversarial one. In fact, the universal rules do not use any information from the method itself (no information from the method, gradient information or training dataset information is used), making them black-box universal attacks. In other words, the universal rules are sample and method agnostic. By proposing a coevolutionary optimization algorithm we show that it is possible to create universal rules that can automatically craft imperceptible adversarial samples (only less than five perturbations which are close to misspelling are inserted in the text sample). A comparison with a random search algorithm further justifies the strength of the method. Thus, universal rules for fooling networks are here shown to exist. Hopefully, the results from this work will impact the development of yet more sample and model agnostic attacks as well as their defenses, culminating in perhaps a new age for artificial intelligence.