Yoshua Bengio, left, has been a machine learning researcher for decades and runs Montreal's MILA institute for AI. Gary Marcus is a psychologist at NYU and a frequent critic of the puffed-up hype around AI. Gary Marcus, the NYU professor and entrepreneur who has made himself a gadfly of deep learning with his frequent skewering of headline hype, and Yoshua Bengio, a leading practitioner of deep learning awarded computing's higher honor for his pioneering work, went head to head Monday night in a two-hour debate Webcast from Bengio's MILA institute headquarters in Montreal. The two scholars seemed to find a lot of common ground as far as the broad strokes of where artificial intelligence needs to go, things such as trying to bring reasoning to AI. But when the discussion periodically lapsed into particular terminology or historical assertions, the two were suddenly at odds. The recorded stream of the video is posted on the organization's Facebook page if you want to go back and watch it.

In this two-day workshop taking place at the National Geospatial-Intelligence Agency (NGA) in Springfield, Va., USGIF and its Machine Learning and Artificial Intelligence Working Group will discuss current challenges and strategic initiatives related to the role of AI, machine learning, cognitive computing, and deep learning. This year's programming features a variety of leading experts from DIA, NGA, NRO, ODNI, OUSD, and industry. Both days are filled with ML AI luminaries from throughout the community including keynotes, panel discussions, exhibits, and flash talks. The first day will be unclassified and the second day will be classified at the TS/SCI/FVEY level.

That may seem a reasonable rate, but many people can't stop there. Serious health and social problems emerge around binge or excessive drinking: illness, crime, traffic accidents, addiction, etc. Alcohol is a factor in 5.3 percent of all deaths worldwide. There is increasing attention on machine learning, deep learning, IoT and computer vision technologies in attempts to reduce the damage done by alcohol and improve the safety of drinkers. AI-powered models can ensure alcohol purity, preventively monitor and assess human behaviours related to drinking, and generate support and services for addicted, intoxicated or unconsciousness people. For generations, Chinese Moatai has been a deeply-loved, top-level distilled liquor.

Applied Brain Research (ABR) announce a new Algorithm that enables advances in ultra-low-power AI speech, vision and signal processing systems for always-on and edge-AI applications, extending battery life while making them more accurate. ABR's announcement demonstrates the potential to realize ultra-low-power instantiations of a large class of algorithms that learn patterns in data, spanning extraordinarily long intervals of time. Current algorithms, like Long Short-Term Memories (LSTMs), can learn and predict sequences of data for long periods of time and make it possible for neural networks to learn to process data like speech, video and control signals. Present in most smart speakers and voice recognition systems, LSTMs are said to be the most financially valuable AI algorithm ever invented (Forbes). LSTMs fail when tasked with learning temporal dependencies in signals than span 1,000 time-steps or more, making them very difficult to scale and limit commercial application.

This repository contains the code and supplementary result for the paper "Unpaired Speech Enhancement by Acoustic and Adversarial Supervision" (IEEE Signal Processing Letters, 2019). We modify the feature from spectrogram to log-Mel filterbank output (LMFB), and 2D convolutional layer to 1D convolutional layer. We trained acoustic model on Librispeech corpus. Generate grapheme probability estimated from the acoustic model on validation data. Find the optimal (alpha, beta) which achieves the lowest WER on the validation data. We search (alpha, beta) by 6x6 grid search.

In the previous article, 'What is Machine Learning?" Deep learning has advanced side-by-side with the digital era, which has led to a massive increase of data in all types. This data, also known as big data, is generated from sources like social media, internet search engines, e-commerce websites, among others. However, this data is normally generated as unstructured and because of the sheer quantity of it, it could take many years to sort and analyse all of it. This is where Deep learning and machine learning come into play. Deep Learning is part of a broader family of machine learning methods based on artificial neural networks. Much like machine learning, deep learning can be supervised, semi-supervised, and unsupervised. Deep learning architectures such as deep belief networks, recurrent neural networks, deep neural networks and convolutional neural networks have been applied to various fields like social network filtering, computer vision, natural language processing, and medical image analysis just to name a few. Machine learning is the most common technique in artificial intelligence. From what we explained in the previous article, machine learning is a self-adaptive algorithm that is continually improved through continual analysis of patterns and new information. Deep learning is a subset of machine learning. Let's delve deeper into the definition of Deep Learning and gain a better understanding of why it has become a subset of machine learning. Artificial intelligence is a set of algorithms and intelligence to try to mimic human intelligence. Machine learning is one of them, and deep learning is one of those machine learning techniques."

In August, Tencent announced it had developed an AI system capable of defeating teams of pros in a five-on-five match in Honor of Kings (or Arena of Valor, depending on the region). This was a noteworthy achievement -- Honor of Kings occupies the video game subgenre known as multiplayer online battle arena games (MOBAs), which are incomplete information games in the sense that players are unaware of the actions other players choose. The endgame, then, isn't merely AI that achieves Honor of Kings superhero performance, but insights that might be used to develop systems capable of solving some of society's toughest challenges. A paper published this week peels back the layers of Tencent's technique, which the coauthors describe as "highly scalable." They claim its novel strategies enable it to explore the game map "efficiently," with an actor-critic architecture that self-improves over time.

To study users' travel behaviour and travel time between origin and destination, researchers employ travel surveys. Although there is consensus in the field about the potential, after over ten years of research and field experimentation, Smartphone-based travel surveys still did not take off to a large scale. Here, computer intelligence algorithms take the role that operators have in Traditional Travel Surveys; since we train each algorithm on data, performances rest on the data quality, thus on the ground truth. Inaccurate validations affect negatively: labels, algorithms' training, travel diaries precision, and therefore data validation, within a very critical loop. Interestingly, boundaries are proven burdensome to push even for Machine Learning methods. To support optimal investment decisions for practitioners, we expose the drivers they should consider when assessing what they need against what they get. This paper highlights and examines the critical aspects of the underlying research and provides some recommendations: (i) from the device perspective, on the main physical limitations; (ii) from the application perspective, the methodological framework deployed for the automatic generation of travel diaries; (iii)from the ground truth perspective, the relationship between user interaction, methods, and data.

In this paper, we study a new graph learning problem: learning to count subgraph isomorphisms. Although the learning based approach is inexact, we are able to generalize to count large patterns and data graphs in polynomial time compared to the exponential time of the original NP-complete problem. Different from other traditional graph learning problems such as node classification and link prediction, subgraph isomorphism counting requires more global inference to oversee the whole graph. To tackle this problem, we propose a dynamic intermedium attention memory network (DIAMNet) which augments different representation learning architectures and iteratively attends pattern and target data graphs to memorize different subgraph isomorphisms for the global counting. We develop both small graphs (<= 1,024 subgraph isomorphisms in each) and large graphs (<= 4,096 subgraph isomorphisms in each) sets to evaluate different models. Experimental results show that learning based subgraph isomorphism counting can help reduce the time complexity with acceptable accuracy. Our DIAMNet can further improve existing representation learning models for this more global problem.

Most optimizers including stochastic gradient descent (SGD) and its adaptive gradient derivatives face the same problem where an effective learning rate during the training is vastly different. A learning rate scheduling, mostly tuned by hand, is usually employed in practice. In this paper, we propose CProp, a gradient scaling method, which acts as a second-level learning rate adapting throughout the training process based on cues from past gradient conformity. When the past gradients agree on direction, CProp keeps the original learning rate. On the contrary, if the gradients do not agree on direction, CProp scales down the gradient proportionally to its uncertainty. Since it works by scaling, it could apply to any existing optimizer extending its learning rate scheduling capability. We put CProp to a series of tests showing significant gain in training speed on both SGD and adaptive gradient method like Adam. Codes are available at https://github.com/phizaz/cprop .