When we are faced with challenging image classification tasks, we often explain our reasoning by dissecting the image, and pointing out prototypical aspects of one class or another. The mounting evidence for each of the classes helps us make our final decision. In this work, we introduce a deep network architecture that reasons in a similar way: the network dissects the image by finding prototypical parts, and combines evidence from the prototypes to make a final classification. The algorithm thus reasons in a way that is qualitatively similar to the way ornithologists, physicians, geologists, architects, and others would explain to people on how to solve challenging image classification tasks. The network uses only image-level labels for training, meaning that there are no labels for parts of images. We demonstrate the method on the CIFAR-10 dataset and 10 classes from the CUB-200-2011 dataset.

Research on question answering with knowledge base has recently seen an increasing use of deep architectures. In this extended abstract, we study the application of the neural machine translation paradigm for question parsing. We employ a sequence-to-sequence model to learn graph patterns in the SPARQL graph query language and their compositions. Instead of inducing the programs through question-answer pairs, we expect a semi-supervised approach, where alignments between questions and queries are built through templates. We argue that the coverage of language utterances can be expanded using late notable works in natural language generation.

In this paper, we study the problem of learning vision-based dynamic manipulation skills using a scalable reinforcement learning approach. We study this problem in the context of grasping, a longstanding challenge in robotic manipulation. In contrast to static learning behaviors that choose a grasp point and then execute the desired grasp, our method enables closed-loop vision-based control, whereby the robot continuously updates its grasp strategy based on the most recent observations to optimize long-horizon grasp success. To that end, we introduce QT-Opt, a scalable self-supervised vision-based reinforcement learning framework that can leverage over 580k real-world grasp attempts to train a deep neural network Q-function with over 1.2M parameters to perform closed-loop, real-world grasping that generalizes to 96% grasp success on unseen objects. Aside from attaining a very high success rate, our method exhibits behaviors that are quite distinct from more standard grasping systems: using only RGB vision-based perception from an over-the-shoulder camera, our method automatically learns regrasping strategies, probes objects to find the most effective grasps, learns to reposition objects and perform other non-prehensile pre-grasp manipulations, and responds dynamically to disturbances and perturbations.

Malicious software, or malware, presents a continuously evolving challenge in computer security. These embedded snippets of code in the form of malicious files or hidden within legitimate files cause a major risk to systems with their ability to run malicious command sequences. Malware authors even use polymorphism to reorder these commands and create several malicious variations. However, if executed in a secure environment, one can perform early malware detection on emulated command sequences. The models presented in this paper leverage this sequential data derived via emulation in order to perform Neural Malware Detection. These models target the core of the malicious operation by learning the presence and pattern of co-occurrence of malicious event actions from within these sequences. Our models can capture entire event sequences and be trained directly using the known target labels. These end-to-end learning models are powered by two commonly used structures - Long Short-Term Memory (LSTM) Networks and Convolutional Neural Networks (CNNs). Previously proposed sequential malware classification models process no more than 200 events. Attackers can evade detection by delaying any malicious activity beyond the beginning of the file. We present specialized models that can handle extremely long sequences while successfully performing malware detection in an efficient way. We present an implementation of the Convoluted Partitioning of Long Sequences approach in order to tackle this vulnerability and operate on long sequences. We present our results on a large dataset consisting of 634,249 file sequences, with extremely long file sequences.

Recent studies on neural architecture search have shown that automatically designed neural networks perform as good as human-designed architectures. While most existing works on neural architecture search aim at finding architectures that optimize for prediction accuracy. These methods may generate complex architectures consuming excessively high energy consumption, which is not suitable for computing environment with limited power budgets. We propose MONAS, a Multi-Objective Neural Architecture Search with novel reward functions that consider both prediction accuracy and power consumption when exploring neural architectures. MONAS effectively explores the design space and searches for architectures satisfying the given requirements. The experimental results demonstrate that the architectures found by MONAS achieve accuracy comparable to or better than the state-of-the-art models, while having better energy efficiency.

A panel of AI experts have been grilled on the impact and importance of artificial general intelligence by the US House of Representatives on Tuesday. The hearing was ominously named "Artificial Intelligence – With Great Power Comes Great Responsibility." Narrow AI for specific tasks has been rapidly advancing and the committee wanted to know how far off artificial general intelligence (AGI), where a system can learn multiple actions and do them better than humans, would be. Greg Brockman, co-founder and CTO at OpenAI, defined AGI as "highly autonomous systems that outperform humans at most economically valuable work". Progress is spearheaded by three areas: data, computation, and algorithms. A recent OpenAI study estimated there had been a 300,000-times increase in the amount of compute used to train AI systems since 2012.

Elon Musk OpenAI's algorithms just crushed a team of experienced players in a complex computer game yesterday For the first time ever, OpenAI, startup backed by Elon Musk, has just won a team of experienced players at the video game "Dota 2" "The ability to learn these kinds of video games at human levels is highly important for the advancement of AI. Indeed, these games more closely approximate the uncertainties and complexity of the real world than games such as chess, which IBM's software mastered in the late 1990s, or Go, which was conquered in 2016 with software created by Google's DeepMind" Read more here: https://lnkd.in/fURiEzA

The Cambrian explosion of deep-learning based startups that started a year or two ago has mostly continued unabated, even though the AI startup market is (arguably) showing signs of finally cooling down. Expectations, round sizes and valuations remain high, but we are certainly past the phase where big Internet companies would snap up very early AI startups at high prices just for the talent. The air is also clearing up a bit and revealing "real" AI startups, versus a number of other companies that were leveraging the hype. Some of the AI startups that were founded in the 2014-2016 time frame are starting to hit early scale, and many are offering increasingly interesting products across industries and verticals including health, finance, "industry 4.0" and back office automation. Deep learning will continue bringing a lot of value in real world applications for years to come, and vertical-focused AI startups have many great opportunities ahead of them.

The field of artificial intelligence continues to grow and evolve daily. To keep up-to-date on all the latest developments, news, and more related to AI, Cognilytica has put together a list of 50 AI influencers to follow on twitter (listed alphabetically). He is also the founder, CEO, and lead designer of SpaceX; co-founder, CEO, and product architect of Tesla, Inc.; and co-founder and CEO of Neuralink. Community leader for Data Is Beautiful 39 @rodneyabrooks – Rodney Brooks, Founder, Chairman, CTO Rethink Robotics.

IN THE MODERN FIELD OF ARTIFICIAL INTELLIGENCE, all roads seem to lead to three researchers with ties to Canadian universities. The first, Geoffrey Hinton, a 70-year-old Brit who teaches at the University of Toronto, pioneered the subfield called deep learning that has become synonymous with A.I. The second, a 57-year-old Frenchman named Yann LeCun, worked in Hinton's lab in the 1980s and now teaches at New York University. The third, 54-year-old Yoshua Bengio, was born in Paris, raised in Montreal, and now teaches at the University of Montreal. The three men are close friends and collaborators, so much so that people in the A.I. community call them the Canadian Mafia. In 2013, though, Google recruited Hinton, and Facebook hired LeCun. Both men kept their academic positions and continued teaching, but Bengio, who had built one of the world's best A.I. programs at the University of Montreal, came to be seen as the last academic purist standing. Bengio is not a natural industrialist. He has a humble, almost apologetic, manner, with the slightly stooped bearing of a man who spends a great deal of time in front of computer screens.