"Children with autism spectrum conditions often have trouble recognizing the emotional states of people around them -- distinguishing a happy face from a fearful face, for instance. To remedy this, some therapists use a kid-friendly robot to demonstrate those emotions and to engage the children in imitating the emotions and responding to them in appropriate ways."

"This type of therapy works best, however, if the robot can smoothly interpret the child's own behavior -- whether he or she is interested and excited or paying attention -- during the therapy. Researchers at the MIT Media Lab have now developed a type of personalized machine learning that helps robots estimate the engagement and interest of each child during these interactions, using data that are unique to that child."

"Armed with this personalized 'deep learning' network, the robots' perception of the children's responses agreed with assessments by human experts, with a correlation score of 60 percent."

"DLTK, the Deep Learning Toolkit for Medical Imaging, extends TensorFlow to enable deep learning on biomedical images." "Biomedical images are typically volumetric images (3D) and sometimes have an additional time dimension (4D) and/or multiple channels (4-5D) (e.g. multi-sequence MR images). The variation in biomedical images is quite different from that of a natural image (e.g. a photograph), as clinical protocols aim to stratify how an image is acquired (e.g. a patient is lying on his/her back, the head is not tilted, etc.). In their analysis, we aim to detect subtle differences (i.e. some small region indicating an abnormal finding)."

"The main reasons for creating DLTK were to include speciality tools for this domain out of the box. While many deep learning libraries expose low-level operations (e.g. tensor multiplications, etc.) to the developers, a lot of the higher-level specialty operations are missing for their use on volumetric images (e.g. differentiable 3D upsampling layers, etc.), and due to the additional spatial dimension(s) of the images, we can run into memory issues (e.g. storing a single copy of a database of 1k CT images, with image dimensions of 512x512x256 voxels in float32 is ~268 GB). Due to the different nature of acquisition, some images will require special pre-processing (e.g. intensity normalization, bias-field correction, de-noising, spatial normalization/registration, etc)."

"The most eye-catching of Volta's new features are the new specialized processing blocks -- tensor cores -- but as we will see, this is very much integrated with the rest of Volta's microarchitectural improvements and surrounding software/framework support for deep learning (DL) and high performance compute (HPC). Matching up with the NVIDIA Titan V are the Titan Xp and GeForce GTX Titan X (Maxwell), with the AMD Radeon RX Vega 64 also present for some tests."

China's first AI chip has been produced by Baidu. The processor is called Kunlun and "is capable of handling both datacentre and edge workloads." "The company said it started developing a field-programmable gate array AI accelerator in 2011, and that Kunlun is almost 30 times faster. The chips are made with Samsung's 14nm process, have 512GBps memory bandwidth, and are capable of 260 tera operations per second at 100 watts."

《L'Italia ha deciso di organizzare le eccellenze che ha già riunendole sotto all'unico cappello del Laboratorio Nazionale di Intelligenza Artificiale AIIS "Artificial Intelligence and Intelligent Systems" promosso dal Consorzio Interuniversitario Nazionale per l'Informatica (CINI).

Vi partecipano 43 atenei e 600 ricercatori del Belpaese, fra cui l'Istituto Italiano di Tecnologia, le Università degli Studi di Milano, di Roma, Palermo, di Bari e di Napoli, e il CNR (Consiglio Nazionale delle Ricerche). Obiettivo è di concentrarsi sull'Intelligenza Artificialee sui settori dove si stima che ci sarà una maggiore influenza, fra cui la medicina, la sicurezza informatica, l'industria automobilistica e la digitalizzazione dei media.》

Very interesting.... 🤖

"The unreasonable effectiveness of deep learning representations." "We build an end-to-end model that is trained on all our images to take an image as an input, and output a similarity score over all of our images. Predictions happen quickly (one forward pass), but we would need to train a new model every time we add a new image. We would also quickly reach a state with so many classes that it would be extremely hard to optimize it correctly."

"Another approach is to build a model that takes in two images, and outputs a pairwise similarity score between 0 and 1. These models are accurate for large datasets, but lead to another scalability issue. We usually want to find a similar image by looking through a vast collection of images, so we have to run our similarity model once for each image pair in our dataset. If our model is a CNN, and we have more than a dozen images, this becomes too slow to even be considered."

"There is a simpler method, which is similar to word embeddings. If we find an expressive vector representation, or embedding for images, we can then calculate their similarity by looking at how close their vectors are to each other."

Are you ready for this new era? (Article in spanish) Hope you find interesting!

https://espabot.es/robots-nos-quitan-trabajo/

The top 5 ways AI will change business: AI, the buzzword of the moment, is applied to everything from speakers to cars. It's hard to separate the hype from the reality sometimes, but here is some reality. From +TechRepublic:

https://www.techrepublic.com/article/top-5-ways-ai-will-change-business/

#software #technology #cloudservices #itsolutions #electronics #compsci #instatech #business #cyberattacks #smashingit