Many types of data from fields including natural language processing, computer vision, and bioinformatics, are well represented by discrete, compositional structures such as trees, sequences, or matchings. Latent structure models are a powerful tool for learning to extract such representations, offering a way to incorporate structural bias, discover insight about the data, and interpret decisions. However, effective training is challenging, as neural networks are typically designed for continuous computation. This text explores three broad strategies for learning with discrete latent structure: continuous relaxation, surrogate gradients, and probabilistic estimation. Our presentation relies on consistent notations for a wide range of models. As such, we reveal many new connections between latent structure learning strategies, showing how most consist of the same small set of fundamental building blocks, but use them differently, leading to substantially different applicability and properties.

In federated learning (FL), it is commonly assumed that all data are placed at clients in the beginning of machine learning (ML) optimization (i.e., offline learning). However, in many real-world applications, it is expected to proceed in an online fashion. To this end, online FL (OFL) has been introduced, which aims at learning a sequence of global models from decentralized streaming data such that the so-called cumulative regret is minimized. Combining online gradient descent and model averaging, in this framework, FedOGD is constructed as the counterpart of FedSGD in FL. While it can enjoy an optimal sublinear regret, FedOGD suffers from heavy communication costs. In this paper, we present a communication-efficient method (named OFedIQ) by means of intermittent transmission (enabled by client subsampling and periodic transmission) and quantization. For the first time, we derive the regret bound that captures the impact of data-heterogeneity and the communication-efficient techniques. Through this, we efficiently optimize the parameters of OFedIQ such as sampling rate, transmission period, and quantization levels. Also, it is proved that the optimized OFedIQ can asymptotically achieve the performance of FedOGD while reducing the communication costs by 99%. Via experiments with real datasets, we demonstrate the effectiveness of the optimized OFedIQ.

Video-Language Pre-training models have recently significantly improved various multi-modal downstream tasks. Previous dominant works mainly adopt contrastive learning to achieve global feature alignment across modalities. However, the local associations between videos and texts are not modeled, restricting the pre-training models' generality, especially for tasks requiring the temporal video boundary for certain query texts. This work introduces a novel text-video localization pre-text task to enable fine-grained temporal and semantic alignment such that the trained model can accurately perceive temporal boundaries in videos given the text description. Specifically, text-video localization consists of moment retrieval, which predicts start and end boundaries in videos given the text description, and text localization which matches the subset of texts with the video features. To produce temporal boundaries, frame features in several videos are manually merged into a long video sequence that interacts with a text sequence. With the localization task, our method connects the fine-grained frame representations with the word representations and implicitly distinguishes representations of different instances in the single modality. Notably, comprehensive experimental results show that our method significantly improves the state-of-the-art performance on various benchmarks, covering text-to-video retrieval, video question answering, video captioning, temporal action localization and temporal moment retrieval. The code will be released soon.

The Artificial Intelligence revolution is here. That might sound like hyperbole. After all, the world looks the same. The revolution didn't arrive with Skynet and robots or with Blade Running cyborgs. A small chat at the bottom right hand corner. If you're imagining Siri or Alexa or even Clippy (Rest in Peace, Clippy), it's so much more than that.

When Coursera CEO Jeff Maggioncalda first "started banging" on OpenAI's ChatGPT, he couldn't believe what he saw. "It looked like magic," he told Insider's Cadie Thompson at the 2023 World Economic Forum. The former English major turned ed-tech executive said that he was impressed by how the buzzy chatbot was able to "recombine word patterns" to "create new ideas." "The first time I sat down in front of ChatGPT, I said'this is not possible,'" Maggioncalda said. He called ChatGPT a "game changer" that is "blowing my mind" -- so much so that he now talks to ChatGPT daily and uses it as a "writing assistant" and a "blog partner." His interest in the AI extends beyond personal use.

This series of courses begins by introducing fundamental Google Cloud concepts to lay the foundation for how businesses use data, machine learning (ML), and artificial intelligence (AI) to transform their business models. The specialization is intended for anyone interested in how the use of AI and ML for the cloud, and especially for data, creates opportunities and requires change for businesses. No previous experience with ML, programming, or cloud technologies is required. The courses do not include any hands-on technical training.

While many Americans were nursing hangovers on New Year's Day, 22-year-old Edward Tian was working feverishly on a new app to combat misuse of a powerful, new artificial intelligence tool called ChatGPT. Given the buzz it's created, there's a good chance you've heard about ChatGPT. It's an interactive chatbot powered by machine learning. The technology has basically devoured the entire Internet, reading the collective works of humanity and learning patterns in language that it can recreate. All you have to do is give it a prompt, and ChatGPT can do an endless array of things: write a story in a particular style, answer a question, explain a concept, compose an email -- write a college essay -- and it will spit out coherent, seemingly human-written text in seconds.

Teachers and students will discover by June that learning "will never be the same again" as people are increasingly using the artificial intelligence chatbot ChatGPT, according to the chief executive of one of the world's largest providers of online education. ChatGPT is able to process large amounts of text and shares that information such as summarising it or explaining it "in a conversational way. The dialogue format makes it possible … to answer follow-up questions, admit its mistakes, challenge incorrect premises and reject inappropriate requests", according to its creators, OpenAI. If asked, it can create essays, poems and coding. Source material can be provided by a user or the technology uses available resources from the internet.

ChatGPT has multiple uses, from writing or fixing a code to getting suggestions, getting explanations, writing non-plagiarised essays, creating summaries of long write-ups, getting solutions to problems etc. The possibilities are still open to explorations as the beta version is available for users to try out. It especially gained popularity when students started realising that they can use ChatGPT to get through their homework assignments and projects by just letting this "assistant" do it for them. Students around the world have been using it to complete their work and teachers have been reporting about how they are doubting the credibility of the work being submitted to them. These developments are catching a lot of traction on the internet.

I wrote about the movement to ban ChatGPT in schools, and why it seems like a missed opportunity. We believe teaching computer science and AI in schools is critically important. Students need to learn how to navigate a world where programming and artificial intelligence are part of our everyday lives.