When it comes to deep learning frameworks, TensorFlow is one of the most preferred toolkits. However, one framework that is fast becoming the favorite of developers and data scientists is PyTorch. PyTorch is an open source project from Facebook which is used extensively within the company. For a long time, Facebook developers used another homegrown framework called Caffe2, which was adopted by academia and researchers. Last year, Facebook announced that it is merging the efforts of developing Caffe2 and PyTorch to focus on creating a unified framework that is accessible to the community.

Artificial Neural Networks are the computational models inspired by the human brain. Many of the recent advancements have been made in the field of Artificial Intelligence, including Voice Recognition, Image Recognition, Robotics using Artificial Neural Networks. These biological methods of computing are considered to be the next major advancement in the Computing Industry. The term'Neural' is derived from the human (animal) nervous system's basic functional unit'neuron' or nerve cells which are present in the brain and other parts of the human (animal) body. It receives signals from other neurons. It sums all the incoming signals to generate input.

The organization created a machine learning algorithm, GPT-2, that can produce natural-looking language largely indistinguishable from that of a human writer while largely "unsupervised" – it needs only a small prompt text to provide the subject and context for the task. The team have made some strides toward this lofty goal, but have also somewhat inadvertently admitted that, once perfected, the device can mass-produce fake news on an unprecedented scale. "We have observed various failure modes," the team observed. "Such as repetitive text, world modelling failures (eg the model sometimes writes about fires happening under water), and unnatural topic switching." Here's a short story i generated using OpenAI's GPT-2 tool (prompt in bold) pic.twitter.com/DGIVwGuAUV

With plenty of machine learning tools currently available, why would you ever choose an artificial neural network over all the rest? This clip and the next could open your eyes to their awesome capabilities! You'll get a closer look at neural nets without any of the math or code - just what they are and how they work. Soon you'll understand why they are such a powerful tool! Deep Learning is primarily about neural networks, where a network is an interconnected web of nodes and edges.

"Sparsity, that's the direction where deep learning should expand," says Gopi Prashanth, who is vice president of engineering at AI-startup Landing AI, run by former Google AI luminary Andrew Ng. In an interview with ZDNet, Prashanth reflected on the challenge of taking something built for really big data, the machine learning approach called deep learning, and re-engineering it for very little data, perhaps just one single sample at a time. It is not an academic concern. The mandate of Ng and his team is to put AI to work for business. That requires using techniques such as machine learning in some settings where there my be very few good examples of a problem to use to train the machine.

Researchers at the non-profit AI research group OpenAI just wanted to train their new text generation software to predict the next word in a sentence. It blew away all of their expectations and was so good at mimicking writing by humans they've decided to pump the brakes on the research while they explore the damage it could do. Elon Musk has been clear that he believes artificial intelligence is the "biggest existential threat" to humanity. Musk is one of the primary funders of OpenAI and though he has taken a backseat role at the organization, its researchers appear to share his concerns about opening a Pandora's box of trouble. This week, OpenAI shared a paper covering their latest work on text generation technology but they're deviating from their standard practice of releasing the full research to the public out of fear that it could be abused by bad actors.

Adversarial examples in machine learning for images are widely publicized and explored. Illustrations of misclassifications caused by slightly perturbed inputs are abundant and commonly known (e.g., a picture of panda imperceptibly perturbed to fool the classifier into incorrectly labeling it as a gibbon). Similar attacks on deep learning (DL) for radio frequency (RF) signals and their mitigation strategies are scarcely addressed in the published work. Yet, RF adversarial examples (AdExs) with minimal waveform perturbations can cause drastic, targeted misclassification results, particularly against spectrum sensing/survey applications (e.g. BPSK is mistaken for 8-PSK). Our research on deep learning AdExs and proposed defense mechanisms are RF-centric, and incorporate physical world, over-the-air (OTA) effects. We herein present defense mechanisms based on pre-training the target classifier using an autoencoder. Our results validate this approach as a viable mitigation method to subvert adversarial attacks against deep learning-based communications and radar sensing systems.

Distributed learning is central for large-scale training ofdeep-learning models. However, they are exposed to a security threat in which Byzantine participants can interrupt or control the learning process. Previous attack models and their corresponding defensesassume that the rogue participants are (a) omniscient (know the data of all other participants), and (b) introduce large change to the parameters. We show that small but wellcrafted changesare sufficient, leading to a novel non-omniscient attack on distributed learning that go undetected by all existing defenses. We demonstrate ourattack method works not only for preventing convergencebut also for repurposing of the model behavior ("backdooring"). We show that 20% of corrupt workers are sufficient to degrade aCIFAR10 model's accuracy by 50%, as well as to introduce backdoors into MNIST and CIFAR10 models without hurting their accuracy.

Sum-Product Networks (SPNs) are hierarchical probabilistic graphical models capable of fast and exact inference. Applications of SPNs to real-world data such as large image datasets has been fairly limited in previous literature. We introduce Convolutional Sum-Product Networks (ConvSPNs) which exploit the inherent structure of images in a way similar to deep convolutional neural networks, optionally with weight sharing. ConvSPNs encode spatial relationships through local products and local sum operations. ConvSPNs obtain state-of-the-art results compared to other SPN-based approaches on several visual datasets, including color images, for both generative as well as discriminative tasks. ConvSPNs are the first pure-SPN models applied to color images that do not depend on additional techniques for feature extraction. In addition, we introduce two novel methods for regularizing SPNs trained with hard EM. Both regularization methods have been motivated by observing an exponentially decreasing variance of log probabilities with respect to the depth of randomly structured SPNs. We show that our regularization provides substantial further improvements in generative visual tasks.

One of the ways to train deep neural networks effectively is to use residual connections. Residual connections can be classified as being either identity connections or bridge-connections with a reshaping convolution. Empirical observations on CIFAR-10 and CIFAR-100 datasets using a baseline Resnet model, with bridge-connections removed, have shown a significant reduction in accuracy. This reduction is due to lack of contribution, in the form of feature maps, by the bridge-connections. Hence bridge-connections are vital for Resnet. However, all feature maps in the bridge-connections are considered to be equally important. In this work, an upgraded architecture "Res-SE-Net" is proposed to further strengthen the contribution from the bridge-connections by quantifying the importance of each feature map and weighting them accordingly using Squeeze-and-Excitation (SE) block. It is demonstrated that Res-SE-Net generalizes much better than Resnet and SE-Resnet on the benchmark CIFAR-10 and CIFAR-100 datasets.