We study a multitask learning problem in which each task is parametrized by a weight vector and indexed by a pair of indices, which can be e.g, (consumer, time). The weight vectors can be collected into a tensor and the (multilinear-)rank of the tensor controls the amount of sharing of information among tasks. Two types of convex relaxations have recently been proposed for the tensor multilinear rank. However, we argue that both of them are not optimal in the context of multitask learning in which the dimensions or multilinear rank are typically heterogeneous. We propose a new norm, which we call the scaled latent trace norm and analyze the excess risk of all the three norms. The results apply to various settings including matrix and tensor completion, multitask learning, and multilinear multitask learning. Both the theory and experiments support the advantage of the new norm when the tensor is not equal-sized and we do not a priori know which mode is low rank.

Deep learning optimizers are often motivated through a mix of convex and approximate second-order theory. We select three such methods -- Adam, Shampoo and Prodigy -- and argue that each method can instead be understood as a squarely first-order method without convexity assumptions. In fact, after switching off exponential moving averages, each method is equivalent to steepest descent under a particular norm. By generalizing this observation, we chart a new design space for training algorithms. Different operator norms should be assigned to different tensors based on the role that the tensor plays within the network. For example, while linear and embedding layers may have the same weight space of $\mathbb{R}^{m\times n}$, these layers play different roles and should be assigned different norms. We hope that this idea of carefully metrizing the neural architecture might lead to more stable, scalable and indeed faster training.

It is likely that 2023 will be remembered as the year of Artificial Intelligence (AI). ChatGPT [2] was the fastest internet service to reach 100million users until now (May 2023) [3] and the technology of Large Language Models (LLMs) at its core is a fundamental element of sister apps for images such as Dall-e2, Midjourney and many others. One of the most fascinating aspects of LLMs is that they exhibit unpredicted emergent features. While the media excitedly reported how AI art generator have developed their own taste [4] or chatbots are able to pass school level exams in a growing number of disciplines [5], only in 2023 it was released that, for the past two years, GPT models had consistently improved its performance in tests designed to measure theory of mind in children [6]. For anyone familiar with complexity science, observing emergent properties in a complex system made of billions of artificial neurons is perhaps not surprising, but the growth in human-, or even superhuman-, like capabilities has attracted huge attention from the media and the public, sparking a hectic debate between the technology apocalyptic and integrated [7]. While it is clear that AI could bring us spectacular benefits, from better medical diagnosing to drug discovering, the risks have so far catalysed most of the public attention. Perils associated to narrow AI include increasing opportunities for manipulation of people, enhancing and dehumanising weapons, and rendering human labour increasingly obsolescent [8]. On the other hand, selfimproving "artificial general intelligence" (AGI) could pose an existential threat to humanity itself.

Siri is known to cut people off midsentence, but there's away to make Siri listen longer. CyberGuy shows you how to customize the wait time. So, you've got some money to invest, and you're mulling over your options, right? That's right, folks, the computer wizards have been busy, and they've conjured up a whopper: Artificial Intelligence (AI) investment platforms. It feels like we've been hurled into a sci-fi saga where the heroes and heroines don't carry lightsabers but wield algorithms and datasets.

Throw out your long-term business plans; they are already outdated as you make them. The best way to be prepared for the coming year is by being proactive, responsive and open to new possibilities around every corner. You may very well have only your intelligence and the key skill of remaining prepared for options. It is a slight exaggeration but going by the changes, we have seen in 2020, reality is wilder than fiction. However, thankfully not everything is unpredictable.

Technology is the future of human lives. Over the years, advancements in this field have continued to reshape the world we live in, and it is not stopping now. On a daily basis, developers are working relentlessly to create new software and algorithms that will transform the way people communicate, treat patients, do business, fight crime, and generally exist. AI is at the core of this technological future. With AI, some aspects of the fantasy-world projected in science-fiction movies will soon become a reality.

We study a multitask learning problem in which each task is parametrized by a weight vector and indexed by a pair of indices, which can be e.g, (consumer, time). The weight vectors can be collected into a tensor and the (multilinear-)rank of the tensor controls the amount of sharing of information among tasks. Two types of convex relaxations have recently been proposed for the tensor multilinear rank. However, we argue that both of them are not optimal in the context of multitask learning in which the dimensions or multilinear rank are typically heterogeneous. We propose a new norm, which we call the scaled latent trace norm and analyze the excess risk of all the three norms. The results apply to various settings including matrix and tensor completion, multitask learning, and multilinear multitask learning.

Imagine: in a few years, consumers will be talking to robots on a regular basis. From answering customer support questions to facilitating voice-based searches, artificial intelligence will become the new norm. Artificial intelligence has already become the new norm. Just take a look at today's popular consumer devices--Siri, Alexa, and even Google's new AI assistant with a voice that is almost indistinguishable from a human's on the phone. In one study of 600 executives in 18 countries, from MIT Technology Review and Genesys, 90% of respondents said that their firms are using AI to improve customer experiences.

Except for the sexism that continues to persist throughout the market, toys have evolved a lot in recent decades. Children are now having fun with new tech-heavy toys -- a trend that has only grown stronger with the digital revolution. Dolls, fire trucks, electric trains, and many other toys are not just "lifeless" miniature replicas of their real-life inspirations like they used to be. For one thing, electronic chips have become omnipresent in toys, adding more and more features, functions, and interactivity to them. But this was only the start of the toy tech revolution.

The move towards an intelligent world is faster and more rapid than it ever was before. The increase in this transition has been propagated through the role of several high key stakeholders that have redefined the way we look at technology. One of the key players in this transition is Huawei. Huawei's recent UBBF conference held in Hangzhou on October 18-19 was a step towards awareness in this regard. Being personally present at this conference, there were numerous intakes that I noted down and would like to present to my readers.