Automatic translation from natural language descriptions into programs is a longstanding challenging problem. In this work, we consider a simple yet important sub-problem: translation from textual descriptions to If-Then programs. We devise a novel neural network architecture for this task which we train end-to-end. Specifically, we introduce Latent Attention, which computes multiplicative weights for the words in the description in a two-stage process with the goal of better leveraging the natural language structures that indicate the relevant parts for predicting program elements. Our architecture reduces the error rate by 28.

Have you ever asked, why can't I just tell the PC what I want it to do? If you're in that camp, there's good news: A new AI update to the Windows Power Automate app may make that future a reality. Microsoft is finally adding Copilot-like capabilities to Power Automate, the "power macro" Windows app that you likely have never used. And one of the best additions will be the ability to literally talk the AI through the process. It's not here yet -- it's part of an early access program -- but this could be an extremely powerful update when it finally arrives.

Automatic translation from natural language descriptions into programs is a longstanding challenging problem. In this work, we consider a simple yet important sub-problem: translation from textual descriptions to If-Then programs. We devise a novel neural network architecture for this task which we train end-toend. Specifically, we introduce Latent Attention, which computes multiplicative weights for the words in the description in a two-stage process with the goal of better leveraging the natural language structures that indicate the relevant parts for predicting program elements. Our architecture reduces the error rate by 28.57% compared to prior art [3]. We also propose a one-shot learning scenario of If-Then program synthesis and simulate it with our existing dataset. We demonstrate a variation on the training procedure for this scenario that outperforms the original procedure, significantly closing the gap to the model trained with all data.

The Internet of Things (IoT), with its ubiquitous sensors and streams of massive data for massive insights, has an estimated market valuation of nearly $2 trillion. Apparently, the "sensoring" of the world is a seriously big deal, generating insights into people, processes, and products on a scale that is almost incomprehensible. Certainly, $2 trillion is almost an incomprehensible figure. The corresponding forecasts for data-driven insights that lead to such a valuation are expected to be on a similarly large scale to justify those astronomical projections. But insights are not hardcoded within Raspberry Pi or Arduino kits, though IFTTT (If-This-Then-That) kits might be a satisfactory solution (more about that later).

A new term in the world of digital marketing is automation. It is a term used to reference activities that help digital marketers drive and increase interaction with their audience. This includes things like email marketing, SMS, online advertising, social media management, etc. This article reviews 5 tools that can be used to automate your digital marketing tasks, as well as gives you an introduction to what this all entails. Axiom.ai is a great choice for digital marketing automation and it can help to improve customer experience, increase website traffic, generate leads, and boost engagement with your customers.

One way to build out a smart home is to buy lots of components--sensors, smart bulbs, security cameras, speakers, and whatnot--and connect them all to a hub that helps them communicate with each other and with you, via your smartphone. But let's be real: That can involve spending a lot of money and investing a lot of time. If your wants and needs are simpler, just a few relatively inexpensive products will deliver most of the conveniences a high-end smart home can deliver, and on a much more modest budget. And if you make sure those smart home products are compatible with each other, you'll build a solid foundation that you can expand over time. The key is knowing which smart home products don't depend on a smart home hub to operate.

IFTTT has brought us plenty of cool tricks -- from allowing your Roomba or TomTom GPS to talk to your smart home devices to integrating with Domino's pizza tracker -- and of course, users have been able to build their own "if this then that" applets. Now, IFTTT is ready to move beyond "if this then that," so it's introducing IFTTT Pro, which will allow users to create applets with queries, conditional logic and multiple actions. For instance, IFTTT says Pro can create an applet that, in the evenings, queries both your Google Calendar and Slack before turning on your Philips Hue lights and playing your evening Spotify playlist. While this is a useful service, IFTTT is making a "big and potentially controversial change." It will charge $9.99 per month for Pro, and subscribers will be able to create unlimited Applets. But its regular, free plan will now limit users to creating just three Applets.

The days of creating as many free IFTTT applets as you want are over, with IFTTT announcing that--like so many other formerly free services--it's going the subscription route. The IFTTT Pro plan, which will allow subscribers to create an unlimited number of applets with new and more advanced features, will ultimately cost $10 a month, although if you sign up before October 7, you'll be able to pay what you want for a year. Meanwhile, IFTTT's "free forever" plans will now cap the number of applets you can create to just three, and if you've already created more than three applets, you'll need to pick which three you want to keep if you're planning on sticking with a free account. IFTTT built a loyal following over the years, with its users building and sharing millions of free applets that connect a wide variety of online services and gadgets, ranging from Google Calendar and Twitter to Roombas and Philips Hue smart bulbs. Free users will still be able to use as many IFTTT community applets as they like, but the new three-applet creation cap poses a dilemma for those who've built a number of IFTTT applets of their own.

IFTTT can tell you when to change your home's air filter: Last month on our IoT Podcast, we mentioned the new Ecobee service that sends you air filters for your HVAC system. Thanks to IFTTT, if you use 3M Filtrete Smart filters, you can set up an IFTTT recipe when it's time to replace that filter, even if you don't have a smart thermostat. You can't set up an automatic order based on the sensor in the filter, but with IFTTT, you could change the color of a light, add a task on your to-do list, or create some other action that tells you it's time to buy a clean filter replacement. The new IFTTT integration is one of several the company debuted this month. Covariant raises $40M to build better AI for robots: Covariant, a robotics startup that was initially born as an academic research project, raised $45 million in Series B funding this week.

Implementing enterprise process automation often requires significant technical expertise and engineering effort. It would be beneficial for non-technical users to be able to describe a business process in natural language and have an intelligent system generate the workflow that can be automatically executed. A building block of process automations are If-Then programs. In the consumer space, sites like IFTTT and Zapier allow users to create automations by defining If-Then programs using a graphical interface. We explore the efficacy of modeling If-Then programs as a sequence learning task. We find Seq2Seq approaches have high potential (performing strongly on the Zapier recipes) and can serve as a promising approach to more complex program synthesis challenges.