Recently, image forensics community has paied attention to the research on the design of effective algorithms based on deep learning technology and facts proved that combining the domain knowledge of image forensics and deep learning would achieve more robust and better performance than the traditional schemes. Instead of improving it, in this paper, the safety of deep learning based methods in the field of image forensics is taken into account. To the best of our knowledge, this is a first work focusing on this topic. Specifically, we experimentally find that the method using deep learning would fail when adding the slight noise into the images (adversarial images). Furthermore, two kinds of strategys are proposed to enforce security of deep learning-based method. Firstly, an extra penalty term to the loss function is added, which is referred to the 2-norm of the gradient of the loss with respect to the input images, and then an novel training method are adopt to train the model by fusing the normal and adversarial images. Experimental results show that the proposed algorithm can achieve good performance even in the case of adversarial images and provide a safety consideration for deep learning-based image forensics

Modern deep neural networks (DNNs) spend a large amount of their execution time computing convolutions. Winograd's minimal algorithm for small convolutions can greatly reduce the number of arithmetic operations. However, a large reduction in floating point (FP) operations in these algorithms can result in significantly reduced FP accuracy of the result. In this paper we propose several methods for reducing the FP error of these algorithms. Minimal convolution algorithms depend on the selection of several numeric \textit{points} that have a large impact on the accuracy of the result. Some points are known to be better than others, but there is no systematic method selecting points for small convolutions. We show that there are a relatively small number of important cases for DNN convolution, that can be searched empirically. We compared both standard and modified versions of the Winograd algorithm. Further, we demonstrate that both the ordering and value of the points is important, and we propose a canonical evaluation ordering that both reduces FP error and the size of the search space based on Huffman coding. We find that good point selections depend on the values of the points themselves and on symmetries between different points. We show that sets of points with symmetric groups give better results. In addition, we explore other methods to reduce FP error, including mixed-precision convolution, and pairwise addition across DNN channels. Using our methods we can significantly reduce FP error for a given Winograd convolution block size, which allows larger block sizes and reduced computation.

Several dihedral angles prediction methods were developed for protein structure prediction and their other applications. However, distribution of predicted angles would not be similar to that of real angles. To address this we employed generative adversarial networks (GAN). Generative adversarial networks are composed of two adversarially trained networks: a discriminator and a generator. A discriminator distinguishes samples from a dataset and generated samples while a generator generates realistic samples. Although the discriminator of GANs is trained to estimate density, GAN model is intractable. On the other hand, noise-contrastive estimation (NCE) was introduced to estimate a normalization constant of an unnormalized statistical model and thus the density function. In this thesis, we introduce noise-contrastive estimation generative adversarial networks (NCE-GAN) which enables explicit density estimation of a GAN model. And a new loss for the generator is proposed. We also propose residue-wise variants of auxiliary classifier GAN (AC-GAN) and Semi-supervised GAN to handle sequence information in a window. In our experiment, the conditional generative adversarial network (C-GAN), AC-GAN and Semi-supervised GAN were compared. And experiments done with improved conditions were invested. We identified a phenomenon of AC-GAN that distribution of its predicted angles is composed of unusual clusters. The distribution of the predicted angles of Semi-supervised GAN was most similar to the Ramachandran plot. We found that adding the output of the NCE as an additional input of the discriminator is helpful to stabilize the training of the GANs and to capture the detailed structures. Adding regression loss and using predicted angles by regression loss only model could improve the conditional generation performance of the C-GAN and AC-GAN.

The susceptibility of deep learning to adversarial attack can be understood in the framework of the Renormalisation Group (RG) and the vulnerability of a specific network may be diagnosed provided the weights in each layer are known. An adversary with access to the inputs and outputs could train a second network to clone these weights and, having identified a weakness, use them to compute the perturbation of the input data which exploits it. However, the RG framework also provides a means to poison the outputs of the network imperceptibly, without affecting their legitimate use, so as to prevent such cloning of its weights and thereby foil the generation of adversarial data.

Let's say I am given an Excel sheet with data about various fruits and I have to tell which look like Apples. What I will do is ask a question "Which fruits are red and round?" and divide all fruits which answer yes and no to the question. Now, All Red and Round fruits might not be apples and all apples won't be red and round. So I will ask a question "Which fruits have red or yellow color hints on them? " on red and round fruits and will ask "Which fruits are green and round?" on not red and round fruits. Based on these questions I can tell with considerable accuracy which are apples. This cascade of questions is what a decision tree is. However, this is a decision tree based on my intuition.

Editor's note: This post is only one part of a far more thorough and in-depth original, found here, which covers much more than what is included here. Working with unstructured text data is hard especially when you are trying to build an intelligent system which interprets and understands free flowing natural language just like humans. You need to be able to process and transform noisy, unstructured textual data into some structured, vectorized formats which can be understood by any machine learning algorithm. Principles from Natural Language Processing, Machine Learning or Deep Learning all of which fall under the broad umbrella of Artificial Intelligence are effective tools of the trade. Based on my previous posts, an important point to remember here is that any machine learning algorithm is based on principles of statistics, math and optimization.

Recently, I've been spending a lot of time thinking about AI. It's understandable because the technology is quickly seeping into every corner of modern life, present in everything from Autonomous Vehicles to I-phone's Siri. As AI automates repetitive tasks, adds intelligence to existing products, achieves impossible accuracy, and adapts through progressive learning it will become the most important technological phenomena of the 22nd century -- second, perhaps, only to the Blockchain. The exact definition of AI is hotly debated and there are already many fantastic explanations of AI on the internet, so I won't dive in too deeply. But broadly speaking, AI is advanced statistics and applied mathematics which harnesses new advances in computing power and the explosion of available data to give computers new powers of inference, recognition, and choice. Machine learning (ML), the most promising subset of AI, is a field that aims to teach computers to learn from examples (or "Data") and perform a task without being explicitly programmed to do so.

Over the last couple years, I've spent an increasing amount of time diving into the possibilities Deep Learning (DL) offers in terms of what we can do with Artificial Intelligence (AI). Some of these possibilities have already been realized (more on this later in the post). And, I could not be more excited to see them out in the world. Through it all, I've felt there are a handful of breath-taking realities that most people are not grasping when it comes to an AI-Powered world. Why the implications are far deeper for humanity than we imagine. Why in my areas of expertise, marketing, sales, customer service and analytics, the impact will be deep and wide. Why is this not yet another programmatic moment. Why the scale at which we can (/have to) solve the problems is already well beyond the grasp of the fundamental strategy most companies follow: We have a bigger revenue opportunity, but we don't know how to take advantage? Let's buy more hamster wheels, hire more hamsters and train them to spin faster!

As usual, NVIDIA CEO Jensen Huang revealed a ton of news during his keynote at the company's GPU Technology Conference yesterday. There's the new Quadro GV100 GPU, which is based on NVIDIA's Volta architecture and will power its new RTX ray tracing technology. The company also revealed its Drive Constellation system for testing self-driving cars in virtual reality, which will certainly help now that it's pausing real world testing. Finally, NVIDIA made some major announcements around AI: its new DGX-2 "personal supercomputer" is insanely powerful, and it's also partnering with ARM to bring its deep learning technology into upcoming Trillium mobile chips.

GridGain Systems, provider of enterprise-grade in-memory computing solutions based on Apache Ignite(TM), today announced the immediate availability of GridGain Professional Edition 2.4, a fully supported version of Apache Ignite 2.4. GridGain Professional Edition 2.4 now includes a Continuous Learning Framework, which includes machine learning and a multilayer perceptron (MLP) neural network that enable companies to run machine and deep learning algorithms against their petabyte-scale operational datasets in real-time. Companies can now build and continuously update models at in-memory speeds and with massive horizontal scalability. GridGain Professional Edition 2.4 also enhances the performance of Apache Spark(TM) by introducing an API for Apache Spark DataFrames, adding to the existing support for Spark RDDs. GridGain Continuous Learning Framework GridGain Professional Edition 2.4 now includes the first fully supported release of the Apache Ignite integrated machine learning and multilayer perceptron features, making continuous learning using machine learning and deep learning available directly in GridGain.