Self-supervised video representation learning aimed at maximizing similarity between different temporal segments of one video, in order to enforce feature persistence over time. This leads to loss of pertinent information related to temporal relationships, rendering actions such as `enter' and `leave' to be indistinguishable. To mitigate this limitation, we propose Latent Time Navigation (LTN), a time-parameterized contrastive learning strategy that is streamlined to capture fine-grained motions. Specifically, we maximize the representation similarity between different video segments from one video, while maintaining their representations time-aware along a subspace of the latent representation code including an orthogonal basis to represent temporal changes. Our extensive experimental analysis suggests that learning video representations by LTN consistently improves performance of action classification in fine-grained and human-oriented tasks (e.g., on Toyota Smarthome dataset). In addition, we demonstrate that our proposed model, when pre-trained on Kinetics-400, generalizes well onto the unseen real world video benchmark datasets UCF101 and HMDB51, achieving state-of-the-art performance in action recognition.

The graph represents a network of 3,954 Twitter users whose tweets in the requested range contained "smarthome", or who were replied to or mentioned in those tweets. The network was obtained from the NodeXL Graph Server on Saturday, 09 July 2022 at 03:37 UTC. The requested start date was Saturday, 09 July 2022 at 00:01 UTC and the maximum number of days (going backward) was 14. The maximum number of tweets collected was 7,500. The tweets in the network were tweeted over the 12-day, 18-hour, 32-minute period from Saturday, 25 June 2022 at 00:06 UTC to Thursday, 07 July 2022 at 18:39 UTC.

The graph represents a network of 3,357 Twitter users whose tweets in the requested range contained "smarthome", or who were replied to or mentioned in those tweets. The network was obtained from the NodeXL Graph Server on Saturday, 28 May 2022 at 03:34 UTC. The requested start date was Saturday, 28 May 2022 at 00:01 UTC and the maximum number of days (going backward) was 14. The maximum number of tweets collected was 7,500. The tweets in the network were tweeted over the 10-day, 8-hour, 57-minute period from Tuesday, 17 May 2022 at 09:45 UTC to Friday, 27 May 2022 at 18:42 UTC.

The graph represents a network of 2,917 Twitter users whose tweets in the requested range contained "smarthome", or who were replied to or mentioned in those tweets. The network was obtained from the NodeXL Graph Server on Saturday, 01 January 2022 at 03:50 UTC. The requested start date was Saturday, 01 January 2022 at 01:01 UTC and the maximum number of days (going backward) was 14. The maximum number of tweets collected was 7,500. The tweets in the network were tweeted over the 10-day, 3-hour, 50-minute period from Tuesday, 21 December 2021 at 14:49 UTC to Friday, 31 December 2021 at 18:40 UTC.

If the invention of the ship was also the invention of the shipwreck, as the French philosopher Paul Virilio suggested, then what does that make the invention of the Nest learning thermostat? As our homes fill up with more connected devices, funnelling every aspect of our lives into the great cloud of big data, the answer could be something much more alarming than just a few more faulty appliances cluttering up our cupboards. This is one of the unsettling questions at the heart of The Future Starts Here, an exhibition about to open at the V&A in London. It promises to be less of a showcase of Tomorrow's World-type gadgetry than a thought-provoking probe into where exactly this new generation of smart technology is taking us. "People seem scared of the future at the moment," says Rory Hyde who, with co-curator Mariana Pestana, has spent the last two years trawling university laboratories and touring Silicon Valley to gather 100 hot-out-of-the-factory innovations, from a low-cost satellite to a solar-powered shirt that can charge a smartphone.