Before the seasonal polar ice cap starts to expand towards lower latitudes on Mars, small frost patches may condensate out during the cold night and they may remain on the surface even during the day in shady areas. If ice in these areas can persist before the arrival of the contiguous ice cap, they may remain after the recession of it too, until the irradiation increases and the ice is met with direct sunlight. In case these small patches form periodically at the same location, slow chemical changes might occur as well. To see the spatial and temporal occurrence of such ice patches, large number of optical images should be searched for and checked. The aim of this study is to survey the ice condensation period on the surface with an automatized method using a Convolutional Neural Network (CNN) applied to High-Resolution Imaging Science Experiment (HiRISE) imagery from the Mars Reconnaissance Orbiter mission. The CNN trained to recognise small ice patches is automatizing the search, making it feasible to analyse large datasets. Previously a manual image analysis was conducted on 110 images from the southern hemisphere, captured by the HiRISE camera. Out of these, 37 images were identified with smaller ice patches, which were used to train the CNN. This approach is applied now to find further images with potential water ice patches in the latitude band between -40{\deg} and -60{\deg}, but contrarily to the training dataset recorded between 140-200{\deg} solar longitude, the images were taken from the condensation period between Ls = 0{\deg} to 90{\deg}. The model was ran on 171 new HiRISE images randomly picked from the given period between -40{\deg} and -60{\deg} latitude band, creating 73155 small image chunks. The model classified 2 images that show small, probably recently condensed frost patches and 327 chunks were predicted to show ice with more than 60% probability.

Digitization of scanned receipts aims to extract text from receipt images and save it into structured documents. This is usually split into two sub-tasks: text localization and optical character recognition (OCR). Most existing OCR models only focus on the cropped text instance images, which require the bounding box information provided by a text region detection model. Introducing an additional detector to identify the text instance images in advance adds complexity, however instance-level OCR models have very low accuracy when processing the whole image for the document-level OCR, such as receipt images containing multiple text lines arranged in various layouts. To this end, we propose a localization-free document-level OCR model for transcribing all the characters in a receipt image into an ordered sequence end-to-end. Specifically, we finetune the pretrained instance-level model TrOCR with randomly cropped image chunks, and gradually increase the image chunk size to generalize the recognition ability from instance images to full-page images. In our experiments on the SROIE receipt OCR dataset, the model finetuned with our strategy achieved 64.4 F1-score and a 22.8% character error rate (CER), respectively, which outperforms the baseline results with 48.5 F1-score and 50.6% CER. The best model, which splits the full image into 15 equally sized chunks, gives 87.8 F1-score and 4.98% CER with minimal additional pre or post-processing of the output. Moreover, the characters in the generated document-level sequences are arranged in the reading order, which is practical for real-world applications.

With neural search seeing rapid adoption, more people are looking at using it for indexing and searching through their unstructured data. I know several folks already building PDF search engines powered by AI, so I figured I'd give it a stab too. How hard could it possibly be? This is just a rough and ready roadmap -- so stay tuned to see how things really pan out. If you want to follow along at home (and maybe fix a few of my bugs!), check the repo: I want to build a search engine for a dataset of arbitrary PDFs.

Climate projections suffer from uncertain equilibrium climate sensitivity. The reason behind this uncertainty is the resolution of global climate models, which is too coarse to resolve key processes such as clouds and convection. These processes are approximated using heuristics in a process called parameterization. The selection of these parameters can be subjective, leading to significant uncertainties in the way clouds are represented in global climate models. Here, we explore three deep network algorithms to infer these parameters in an objective and data-driven way. We compare the performance of a fully-connected network, a one-dimensional and, a two-dimensional convolutional networks to recover the underlying parameters of the Lorenz-96 model, a non-linear dynamical system that has similar behavior to the climate system.

We frequently hear about Machine Learning in the media, especially since the recent wave of interest in deep-learning. The perpetual improvement of Machine Learning techniques combined with the ever increasing amount of data that are stored suggests endless new applications. Many innovative solutions emerge: autonomous driving, next generation supermarkets with implicit payment, next generation chatbots that can interact with you as human beings would do, and so on. More than ever, the future seems within reach. But the more extravagant and original the application is, the more the layman is put off.

We frequently hear about Machine Learning in the media, especially since the recent wave of interest in deep-learning. The perpetual improvement of Machine Learning techniques combined with the ever increasing amount of data that are stored suggests endless new applications. Many innovative solutions emerge: autonomous driving, next generation supermarkets with implicit payment, next generation chatbots that can interact with you as human beings would do, and so on. More than ever, the future seems within reach. But the more extravagant and original the application is, the more the layman is put off.