In personalized recommendation systems, it is important to predict preferences of a user on items that have not been seen by that user yet. Similarly, in revenue management, it is important to predict outcomes of comparisons among those items that have never been compared so far. The MultiNomial Logit model, a popular discrete choice model, captures the structure of the hidden preferences with a low-rank matrix. In order to predict the preferences, we want to learn the underlying model from noisy observations of the low-rank matrix, collected as revealed preferences in various forms of ordinal data. A natural approach to learn such a model is to solve a convex relaxation of nuclear norm minimization. We present the convex relaxation approach in two contexts of interest: collaborative ranking and bundled choice modeling. In both cases, we show that the convex relaxation is minimax optimal. We prove an upper bound on the resulting error with finite samples, and provide a matching information-theoretic lower bound.

Abstract: Recommendation models that utilize unique identities (IDs) to represent distinct users and items have been state-of-the-art (SOTA) and dominated the recommender systems (RS) literature for over a decade. Meanwhile, the pre-trained modality encoders, such as BERT and ViT, have become increasingly powerful in modeling the raw modality features of an item, such as text and images. Given this, a natural question arises: can a purely modality-based recommendation model (MoRec) outperforms or matches a pure ID-based model (IDRec) by replacing the itemID embedding with a SOTA modality encoder? In fact, this question was answered ten years ago when IDRec beats MoRec by a strong margin in both recommendation accuracy and efficiency. We aim to revisit this old' question and systematically study MoRec from several aspects.

Human thinking is beyond imagination. Can a computer develop such ability to think and reason without human intervention? This is something programming experts at IBM Watson are trying to achieve. Their goal is to simulate human thought process in a computerized model. The result is cognitive computing – a combination of cognitive science and computer science. Cognitive computing models provide a realistic roadmap to achieve artificial intelligence.

Woahhh… Surprised that I'm gonna cover every facet of Artificial Intelligence in just a single blog post!? That's definitely impossible!! But, I'm sure I can give you a fundamental grasp of it through this blog. Most of our routines start with unlocking a phone using the fingerprint or facial unlock options and eventually ends up with, "Amazing!! I made it to 10,000 steps today" or "Hey Siri, set the alarm for 5 a.m." Whether we like it or not, we spend a significant amount of time interacting with smart systems, and it's (AI) becoming an essential part of our modern existence. From Search engines to Virtual Assistants, Recommender systems, Google maps, smart homes so on.. Using mathematics and algorithmic techniques, AI solves these complex real-world problems. Artificial Intelligence is a science that develops theories and methodologies to make machines that are capable of thinking and understanding the world intelligently, as well as reacting appropriately to the situation in the same way humans can do.

Since clicks usually contain heavy noise, increasing research efforts have been devoted to modeling implicit negative user behaviors (i.e., non-clicks). However, they either rely on explicit negative user behaviors (e.g., dislikes) or simply treat non-clicks as negative feedback, failing to learn negative user interests comprehensively. In such situations, users may experience fatigue because of seeing too many similar recommendations. In this paper, we propose Fatigue-Aware Network (FAN), a novel CTR model that directly perceives user fatigue from non-clicks. Specifically, we first apply Fourier Transformation to the time series generated from non-clicks, obtaining its frequency spectrum which contains comprehensive information about user fatigue. Then the frequency spectrum is modulated by category information of the target item to model the bias that both the upper bound of fatigue and users' patience is different for different categories. Moreover, a gating network is adopted to model the confidence of user fatigue and an auxiliary task is designed to guide the learning of user fatigue, so we can obtain a well-learned fatigue representation and combine it with user interests for the final CTR prediction. Experimental results on real-world datasets validate the superiority of FAN and online A/B tests also show FAN outperforms representative CTR models significantly.

Graph collaborative filtering (GCF) is a popular technique for capturing high-order collaborative signals in recommendation systems. However, GCF's bipartite adjacency matrix, which defines the neighbors being aggregated based on user-item interactions, can be noisy for users/items with abundant interactions and insufficient for users/items with scarce interactions. Additionally, the adjacency matrix ignores user-user and item-item correlations, which can limit the scope of beneficial neighbors being aggregated. In this work, we propose a new graph adjacency matrix that incorporates user-user and item-item correlations, as well as a properly designed user-item interaction matrix that balances the number of interactions across all users. To achieve this, we pre-train a graph-based recommendation method to obtain users/items embeddings, and then enhance the user-item interaction matrix via top-K sampling. We also augment the symmetric user-user and item-item correlation components to the adjacency matrix. Our experiments demonstrate that the enhanced user-item interaction matrix with improved neighbors and lower density leads to significant benefits in graph-based recommendation. Moreover, we show that the inclusion of user-user and item-item correlations can improve recommendations for users with both abundant and insufficient interactions. The code is in \url{https://github.com/zfan20/GraphDA}.

Startups are key to solving today's biggest challenges and a huge driver of innovation -- and artificial intelligence is one of their sharpest tools. Virtual assistants, customized content, traffic apps, spell check, mobile check deposit and live captioning constitute just a small fraction of the everyday solutions using AI -- and many of these technologies were first developed by startups. AI learns from those who build it, so it is critical to have people of all backgrounds helping shape the technology to ensure its effectiveness, reduce bias and create better solutions for everyone. As Director of Product Inclusion and Equity at Google, I love to see Black founders tap into the power of our Google AI tech to help their communities and transform the way our products work and operate. In honor of Black History Month in the U.S., I asked four Google for Startups Black Founders Fund recipients from around the world and across different industries how they're using Google AI technology to address societal challenges.

Your smartphone is not listening to you around the clock -- but it's collecting so much information that it does not even need to. It has long been speculated that Apple, Google, Samsung and other popular phone makers are recording users 24/7 to collect information for advertising purposes. Most of us have seemingly randomly been promoted an advert for a product that we could have sworn was only talked about in private. To test this, we set up a freshly-factory-reset Samsung phone, using a new Google account on the Android device. We created a fictitious person named Robin, 22, and made a fake a Facebook account for him to use.

Artificial intelligence (AI) is changing the way we live and work, and its impact is only going to grow in the coming years. Here are 10 ways in which AI is transforming our lives: Personal Assistants: AI-powered personal assistants like Siri and Alexa are already popular, and their capabilities are expanding rapidly. These assistants can help us with everything from setting reminders to ordering groceries. Healthcare: AI is being used in healthcare to improve diagnosis, predict outcomes, and develop new treatments. AI-powered medical imaging can detect tumors and other abnormalities, and AI algorithms can predict patient outcomes and identify high-risk patients.

The use of pretrained embeddings has become widespread in modern e-commerce machine learning (ML) systems. In practice, however, we have encountered several key issues when using pretrained embedding in a real-world production system, many of which cannot be fully explained by current knowledge. Unfortunately, we find that there is a lack of a thorough understanding of how pre-trained embeddings work, especially their intrinsic properties and interactions with downstream tasks. Consequently, it becomes challenging to make interactive and scalable decisions regarding the use of pre-trained embeddings in practice. Our investigation leads to two significant discoveries about using pretrained embeddings in e-commerce applications. Firstly, we find that the design of the pretraining and downstream models, particularly how they encode and decode information via embedding vectors, can have a profound impact. Secondly, we establish a principled perspective of pre-trained embeddings via the lens of kernel analysis, which can be used to evaluate their predictability, interactively and scalably. These findings help to address the practical challenges we faced and offer valuable guidance for successful adoption of pretrained embeddings in real-world production. Our conclusions are backed by solid theoretical reasoning, benchmark experiments, as well as online testings.