Our Innovation Analysts recently looked into emerging technologies and up-and-coming startups working on solutions for the railway industry. As there is a large number of startups working on a wide variety of solutions, we decided to share our insights with you. This time, we are taking a look at 4 promising artificial intelligence startups. For our 4 top picks, we used a data-driven startup scouting approach to identify the most relevant solutions globally. The Global Startup Heat Map below highlights 4 interesting examples out of 21 relevant solutions.
OSAKA – Osaka Metro Co. showed a next-generation automated ticket gate with a facial recognition system to the media Monday. Aiming to introduce such gates at all of its train stations in fiscal 2024, ahead of the 2025 World Expo in the city of Osaka, the subway operator will start testing the gates Tuesday with some 1,200 employees. Through the test, the Osaka-based company hopes to identify problems and make improvements. This will be the first such test by a Japanese railway operator, according to Osaka Metro. The test, which is set to run through September 2020, will be conducted at four stations: Dome-mae Chiyozaki, Morinomiya, Dobutsuen-mae and Daikokucho.
We show that the double descent phenomenon occurs in CNNs, ResNets, and transformers: performance first improves, then gets worse, and then improves again with increasing model size, data size, or training time. This effect is often avoided through careful regularization. While this behavior appears to be fairly universal, we don't yet fully understand why it happens, and view further study of this phenomenon as an important research direction. The peak occurs predictably at a "critical regime," where the models are barely able to fit the training set. As we increase the number of parameters in a neural network, the test error initially decreases, increases, and, just as the model is able to fit the train set, undergoes a second descent.
The paper addresses the classical network tomography problem of inferring local traffic given origin-destination observations. Focussing on large complex public transportation systems, we build a scalable model that exploits input-output information to estimate the unobserved link/station loads and the users path preferences. Based on the reconstruction of the users' travel time distribution, the model is flexible enough to capture possible different path-choice strategies and correlations between users travelling on similar paths at similar times. The corresponding likelihood function is intractable for medium or large-scale networks and we propose two distinct strategies, namely the exact maximum-likelihood inference of an approximate but tractable model and the variational inference of the original intractable model. As an application of our approach, we consider the emblematic case of the London Underground network, where a tap-in/tap-out system tracks the start/exit time and location of all journeys in a day.
"Anything that could give rise to smarter-than-human intelligence--in the form of Artificial Intelligence, brain-computer interfaces, or neuroscience-based human intelligence enhancement – wins hands down beyond contest as doing the most to change the world. Nothing else is even in the same league." "Nobody phrases it this way, but I think that artificial intelligence is almost a humanities discipline. It's really an attempt to understand human intelligence and human cognition." "The promise of artificial intelligence and computer science generally vastly outweighs the impact it could have on some jobs in the same way that, while the invention of the airplane negatively affected the railroad industry, it opened a much wider door to human progress."
Toshiba Digital and Consulting Corporation and Mitsui have tested a digital twin software on the route between London's Stansted Airport to London Liverpool Street station. It was in 2002, while at the University of Michigan, that Dr Michael Grieves wrote about bridging the gap between the virtual and real worlds using digital replicas of physical assets, processes or systems. Two decades on, his concept of a'digital twin' has the potential to revolutionise industry. Digital twins use sensors to gather data in real time, which is then processed in a cloud-based system before being compared with other business and contextual data. The resulting analysis enables the operator to predict problems, optimise critical processes, and drive innovation and performance.
Humans can easily detect a defect (anomaly) because it is different or salient when compared to the surface it resides on. Today, manual human visual inspection is still the norm because it is difficult to automate anomaly detection. Neural networks are a useful tool that can teach a machine to find defects. However, they require a lot of training examples to learn what a defect is and it is tedious and expensive to get these samples. We tackle the problem of teaching a network with a low number of training samples with a system we call AnoNet. AnoNet's architecture is similar to CompactCNN with the exceptions that (1) it is a fully convolutional network and does not use strided convolution; (2) it is shallow and compact which minimizes over-fitting by design; (3) the compact design constrains the size of intermediate features which allows training to be done without image downsizing; (4) the model footprint is low making it suitable for edge computation; and (5) the anomaly can be detected and localized despite the weak labelling. AnoNet learns to detect the underlying shape of the anomalies despite the weak annotation as well as preserves the spatial localization of the anomaly. Pre-seeding AnoNet with an engineered filter bank initialization technique reduces the total samples required for training and also achieves state-of-the-art performance. Compared to the CompactCNN, AnoNet achieved a massive 94% reduction of network parameters from 1.13 million to 64 thousand parameters. Experiments were conducted on four data-sets and results were compared against CompactCNN and DeepLabv3. AnoNet improved the performance on an average across all data-sets by 106% to an F1 score of 0.98 and by 13% to an AUROC value of 0.942. AnoNet can learn from a limited number of images. For one of the data-sets, AnoNet learnt to detect anomalies after a single pass through just 53 training images.
Try Hyperloop, rocket travel, and robotic avatars. Hyperloop is currently working towards 670 mph (1080 kph) passenger pods, capable of zipping us from Los Angeles to downtown Las Vegas in under 30 minutes. Rocket Travel (think SpaceX's Starship) promises to deliver you almost anywhere on the planet in under an hour. Think New York to Shanghai in 39 minutes. As 5G connectivity, hyper-realistic virtual reality, and next-gen robotics continue their exponential progress, the emergence of "robotic avatars" will all but nullify the concept of distance, replacing human travel with immediate remote telepresence.
CAIRO - 21 November 2019: The Cabinet, during its meeting on Thursday under Prime Minister Mostafa Madbouli, approved a draft resolution on establishing a national council for artificial intelligence. The national artificial intelligence council, which follows the Cabinet, will be chaired by the communications and information technology minister and group a number of ministers and heads of several bodies concerned. The new national body will be responsible for outlining the national strategy for artificial intelligence and overseeing its implementation in a way that copes up with the international developments in this field. The council will be authorized to cooperate with the related regional and international bodies as well as to select the best artificial intelligence applications that could help offer safe, sustainable and smart services. During the meeting, the Cabinet approved authorizing the ICT minister to contract and sign agreements with Microsoft, ESRI, VMware and Teradata on the behalf of the government to be self-funded by the ministry during the years 2019/2020, 2020/2021, 2021/2022 and 2022/2023.
Facial recognition technology is used across China for everything from identifying criminals to measuring students' attention in class. Now, it has debuted a system in its subway that lets you use your face as a ticket. A report from South China Morning Post suggests the subway system in the southern city of Shenzhen has started using facial recognition technology to let folks over 60 years of age register themselves for free subway rides. Other cities such as Jinan, Shanghai, Qingdao, Nanjing, and Nanning are currently experimenting with this system. The technology in Shenzen has been deployed to 18 stations with 28 automatic gate machines and 60 self-service ticket processors.