We present an approach to map utterances in conversation to logical forms, which will be executed on a large-scale knowledge base. To handle enormous ellipsis phenomena in conversation, we introduce dialog memory management to manipulate historical entities, predicates, and logical forms when inferring the logical form of current utterances. Dialog memory management is embodied in a generative model, in which a logical form is interpreted in a top-down manner following a small and flexible grammar. We learn the model from denotations without explicit annotation of logical forms, and evaluate it on a large-scale dataset consisting of 200K dialogs over 12.8M entities. Results verify the benefits of modeling dialog memory, and show that our semantic parsing-based approach outperforms a memory network based encoder-decoder model by a huge margin.

How did you arrive at this research area?

As the legal war over how to regulate prediction markets rages on, financial institutions are embracing the industry anyway. When Troy Dixon first suggested incorporating prediction markets into the electronic trading platform where he works, he was met with incredulity. "People told us we were crazy," Dixon, Tradeweb's cohead of global markets, tells WIRED. But after the company announced it was partnering with Kalshi in February, Dixon says, the mood changed dramatically. "We've been inundated with calls," he says.

'100 Video Calls Per Day': Models Are Applying to Be the Face of AI Scams Dozens of Telegram channels reviewed by WIRED include job listings for "AI face models." The (mostly) women who land these gigs are likely being used to dupe victims out of their money. "I can speak fluent English, I can speak good Chinese, I also speak Russian and Turkish," the glamorous, 24-year-old Uzbekistani woman explains in a selfie-style video made for recruiters. Angel had arrived in the Cambodian city of Sihanoukville that day, she said, and was ready to start work immediately. Those impressive language skills, however, have likely been put to use as part of elaborate " pig-butchering " scams targeting Americans.

Could Iran be using China's BeiDou system? The United States and Israel continue to carry out attacks as the war with Iran enters its 17th day. Iran has retaliated by launching missile and drone barrages towards Israel and by attacking neighbouring Gulf countries. Unlike the June 2025 strikes, which US President Donald Trump said curbed Iran's nuclear capabilities, the current conflict has spread across at least a dozen countries, closed the Strait of Hormuz - the world's major oil artery - and killed more than 2,300 people in the region. Al Jazeera tracks how events have panned out over the past 16 days.

Neural Information Processing Systems http://nips.cc/

Could Iran be using China's BeiDou system? Iran war: What is happening on day 17 of US-Israel attacks? Israel launched a new wave of attacks on Tehran as the US-Israel war on Iran entered its 17th day on Monday. Escalations continue in the Gulf region, where authorities suspended flights at Dubai international airport after a drone incident sparked a fire nearby. Dubai-based Emirates announced later that it was resuming limited flights, with several planned routes cancelled for the day.

Could Iran be using China's BeiDou system? The United Arab Emirates (UAE) has announced diversion of some flights from the Dubai international airport, one of the world's busiest, after a drone attack sparked a fire near the facility, as Bahrain, Kuwait, Qatar and Saudi Arabia also reported intercepting drones and missiles. The Dubai Media Office on Monday said civil defence teams had "successfully contained the fire resulting from impact to one of the fuel tanks in the vicinity" of the airport, noting that no injuries had been reported so far. The Dubai Civil Aviation Authority, meanwhile, said it was temporarily suspending flights at the airport "as a precautionary measure to ensure the safety of all passengers and staff". It did not say when they expected flights to resume.

We propose a new type of generative model for high-dimensional data that learns a manifold geometry of the data, rather than density, and can generate points evenly along this manifold. This is in contrast to existing generative models that represent data density, and are strongly affected by noise and other artifacts of data collection. We demonstrate how this approach corrects sampling biases and artifacts, thus improves several downstream data analysis tasks, such as clustering and classification. Finally, we demonstrate that this approach is especially useful in biology where, despite the advent of single-cell technologies, rare subpopulations and gene-interaction relationships are affected by biased sampling. We show that SUGAR can generate hypothetical populations, and it is able to reveal intrinsic patterns and mutual-information relationships between genes on a single-cell RNA sequencing dataset of hematopoiesis.

Neural Information Processing Systems http://nips.cc/