Knowledge tracing is the task of modeling each student's mastery of knowledge concepts (KCs) as (s)he engages with a sequence of learning activities. Each student's knowledge is modeled by estimating the performance of the student on the learning activities. It is an important research area for providing a personalized learning platform to students. In recent years, methods based on Recurrent Neural Networks (RNN) such as Deep Knowledge Tracing (DKT) and Dynamic Key-Value Memory Network (DKVMN) outperformed all the traditional methods because of their ability to capture complex representation of human learning. However, these methods face the issue of not generalizing well while dealing with sparse data which is the case with real-world data as students interact with few KCs. In order to address this issue, we develop an approach that identifies the KCs from the student's past activities that are \textit{relevant} to the given KC and predicts his/her mastery based on the relatively few KCs that it picked. Since predictions are made based on relatively few past activities, it handles the data sparsity problem better than the methods based on RNN. For identifying the relevance between the KCs, we propose a self-attention based approach, Self Attentive Knowledge Tracing (SAKT). Extensive experimentation on a variety of real-world dataset shows that our model outperforms the state-of-the-art models for knowledge tracing, improving AUC by 4.43% on average.

For some, a rose is a symbol of beauty or love. For Shaza Mehdi, it is a connection to her mother, but also a gateway to innovation. Mehdi's mother, Afshin, grows rose bushes at their Lawrenceville home. But a few years ago, the plants kept getting diseases, ruining the blooms. Mehdi tried diagnosing the flowers by Googling images of plant diseases and comparing those images with the sick roses.

Sitting in a hotel lobby in Tangier, Morocco, Charity Wayua laughs as she recounts her journey to the city for a conference on technology and innovation. After starting her trip in Nairobi, Kenya, where she leads one of IBM's two research centers in Africa, she had to fly past her destination for a layover in Dubai, double back to Casablanca, and then take a three-and-a-half-hour drive to Tangier. What would have been a seven- to eight-hour direct flight was instead a nearly 24-hour odyssey. This is not unusual, she says. The hassle of traveling within the region isn't the only thing making things difficult for Africa's research community: the difficulty of traveling out of the region has often left its researchers out of the international conversation.

Dustin is posing as an active shooter armed with an assault rifle. If he thinks he's undetected looking to prey on the unsuspecting, he'd be completely wrong. We've tested a couple different model architectures and we use that over existing security cameras using different types of GPUs to be able to digest those video feeds, run analytics over it looking for a weapon and then sending the alert out," said Mike Lahiff, CEO of ZeroEyes. The alert goes out in a flash to law enforcers and administrators with video of Dustin's movements and location. "Instantly, I would get on my police radio and notify first responders that I have a possible threat on location.

AI is finding its way to more places in organizations, including human resources. Human capital management providers are building AI into their solutions, but depending on the details, it may be wiser to build your own application than buy something off-the-shelf. Earlier this year, Gartner issued a research note exploring AI use cases in human capital management (HCM). Its author, VP Analyst Helen Poitevin, concluded that many of these applications were still in the "demo candy" stage, mainly to demonstrate product roadmaps. In other words, AI-related expectations are outpacing reality.

One of Ayush Alag's earliest memories is of biting into a chocolate bar with cashew nuts and suddenly feeling his throat get itchy. For most of his childhood, the Santa Clara, California resident avoided eating anything with cashews and other nuts that caused irritation as best as he could. By his middle school years, he and his parents wanted to know for sure: did he have a serious food allergy, like 32 million other Americans, or was it just a food sensitivity? They sought the help of an allergist, Joseph Hernandez of Stanford University. Hernandez told them that the difference between an allergy and a food sensitivity is huge.

From TED Talk speakers, to a futurologist's keynote at an event, those who make predictions about the future usually live safe in the knowledge they won't retrospectively be pulled up on forecasts that don't come to pass. The picture is very different for those in government, who must ensure citizens and businesses are adequately prepared for challenges. Government predictions must convert to real-world planning that puts building blocks for future success and prosperity in place – it can't be a'finger in the air'. Education is at the foundations of preparation. Governments and educators must identify trends early enough to update curriculums, develop the right courses, and equip people with skills that put us in a strong position to compete on the world stage.

Data science is the application of statistics, programming and domain knowledge to generate insights into a problem that needs to be solved. The Harvard Business Review said Data Scientist is the sexiest job of the 21st century. How often has that article been referenced to convince people? The job'Data Scientist' has been around for decades, it was just not called "Data Scientist". Statisticians have used their knowledge and skills using machine learning techniques such as Logistic Regression and Random Forest for prediction and insights for decades.

One of the challenges in display advertising is that the distribution of features and click through rate (CTR) can exhibit large shifts over time due to seasonality, changes to ad campaigns and other factors. The predominant strategy to keep up with these shifts is to train predictive models continuously, on fresh data, in order to prevent them from becoming stale. However, in many ad systems positive labels are only observed after a possibly long and random delay. These delayed labels pose a challenge to data freshness in continuous training: fresh data may not have complete label information at the time they are ingested by the training algorithm. Naive strategies which consider any data point a negative example until a positive label becomes available tend to underestimate CTR, resulting in inferior user experience and suboptimal performance for advertisers. The focus of this paper is to identify the best combination of loss functions and models that enable large-scale learning from a continuous stream of data in the presence of delayed labels. In this work, we compare 5 different loss functions, 3 of them applied to this problem for the first time. We benchmark their performance in offline settings on both public and proprietary datasets in conjunction with shallow and deep model architectures. We also discuss the engineering cost associated with implementing each loss function in a production environment. Finally, we carried out online experiments with the top performing methods, in order to validate their performance in a continuous training scheme. While training on 668 million in-house data points offline, our proposed methods outperform previous state-of-the-art by 3% relative cross entropy (RCE). During online experiments, we observed 55% gain in revenue per thousand requests (RPMq) against naive log loss.

One of the main, long-term objectives of artificial intelligence is the creation of thinking machines. To that end, substantial effort has been placed into designing cognitive systems; i.e. systems that can manipulate semantic-level information. A substantial part of that effort is oriented towards designing the mathematical machinery underlying cognition in a way that is very efficiently implementable in hardware. In this work we propose a 'semi-holographic' representation system that can be implemented in hardware using only multiplexing and addition operations, thus avoiding the need for expensive multiplication. The resulting architecture can be readily constructed by recycling standard microprocessor elements and is capable of performing two key mathematical operations frequently used in cognition, superposition and binding, within a budget of below 6 pJ for 64- bit operands. Our proposed 'cognitive processing unit' (CoPU) is intended as just one (albeit crucial) part of much larger cognitive systems where artificial neural networks of all kinds and associative memories work in concord to give rise to intelligence.