Some Image and Video Processing: Motion Estimation with Block-Matching in Videos, Noisy and Motion-blurred Image Restoration with Inverse Filter in Python and OpenCV


The following figure shows how the quality of the transformed image decreases when compared to the original image, when an nxn LPF is applied and how the quality (measured in terms of PSNR) degrades as n (LPF kernel width) increases. As we go on increasing the kernel size, the quality fo the final image obtained by down/up sampling the original image decreases as n increases, as shown in the following figure. The first one is the video of some students working on a university corridor, as shown below (obtained from youtube), extract some consecutive frames, mark a face in one image and use that image to mark all thew faces om the remaining frames that are consecutive to each other, thereby mark the entire video and estimate the motion using the simple block matching technique only. The following figure shows the frame with the face marked, now we shall use this image and block matching technique to estimate the motion of the student in the video, by marking his face in all the consecutive frames and reconstructing the video, as shown below.. As can be seen from the following figure, the optimal median filter size is 5 5, which generates the highest quality output, when compared to the original image.

Data Virtualization: Unlocking Data for AI and Machine Learning


Hybrid Execution allows you to "push" queries to a remote system, such as to SQL Server, and access the referential data. However, one can imagine a use case where lots of ETL processing happens in HDInsight clusters and the structured results are published to SQL Server for downstream consumption (for instance, by reporting tools). Note the linear increase in execution time with SQL Server only (blue line) versus when HDInsight is used with SQL Server to scale out the query execution (orange and grey lines). With much larger real-world datasets in SQL Server, which typically runs multiple queries competing for resources, more dramatic performance gains can be expected.

Help EFF Track the Progress of AI and Machine Learning


The field of machine learning and artificial intelligence is making rapid progress. What kinds of problems have been well solved by current machine learning techniques, which ones are close to being solved, and which ones remain exceptionally hard? Today, we're launching the EFF AI Progress Measurement experiment, and encouraging machine learning researchers to give us feedback and contribute to the effort. Given that machine learning tools and AI techniques are increasingly part of our everyday lives, it is critical that journalists, policy makers, and technology users understand the state of the field.


MIT News

When applied to previously-collected atmospheric samples and data, their findings support evidence that on average these bioaerosols globally make up less than 1 percent of the particles in the upper troposphere -- where they could influence cloud formation and by extension, the climate -- and not around 25 to 50 percent as some previous research suggests. While atmospheric and climate modeling suggests that bioaerosols, globally averaged, are not abundant and efficient enough at freezing to significantly influence cloud formation, research findings have varied significantly. The group leveraged the presence of phosphorus in the mass spectra to train the classification machine learning algorithm on known samples and then, primed, applied it to field data acquired from Desert Research Institute's Storm Peak Laboratory in Steamboat Springs, Colorado, and from the Carbonaceous Aerosol and Radiative Effects Study based in the town of Cool, California. Knowing that the principal atmospheric emissions of phosphorus are from mineral dust, combustion products, and biological particles, they exploited the presence of phosphate and organic nitrogen ions and their characteristic ratios in known samples to classify the particles.

Yes, for AI, it really is all about the data


While science fiction often portrays AI as robots with humanlike characteristics, AI can encompass anything from Google's search algorithms to IBM's Watson to autonomous weapons. The big data infrastructure, the deep learning models, and everything else exist to serve the data, not the other way around. AI provides the large-scale analytics needed to extract meaning and benefit from big data, while big data provides the knowledge needed for AI to continue to learn and evolve. Its investments in AI and big data infrastructure are essentially a means to get you to share more data, and then analyze it to sell ads against it.

Python vs R: 4 Implementations of Same Machine Learning Technique


Actually, this is about two R versions (standard and improved), a Python version, and a Perl version of a new machine learning technique recently published here. We asked for help to translate the original Perl script to Python and R, and finally decided to work with Naveenkumar Ramaraju, who is currently pursuing a master's in Data Science at Indiana University. We believe that this code comparison and translation will be very valuable to anyone learning Python or R with the purpose of applying it to data science and machine learning. The code, originally written in Perl, was translated to Python and R by Naveenkumar Ramaraju.

Artificial Intelligence: The new Kid in Town – Insurers.AI – Medium


Apart from the likes of Google, Facebook, Amazon, and Tesla and their mainly digital business models and obvious applications of AI, a lot of traditional industries are employing intelligent algorithms to augment previously manual approaches. AI gives us means to automate processes, personalize products, communications and care, predict personal and collective developments, discover trends and unusual patterns in the data, and more. It has the potential to impact the insurance industry in numerous areas, such as marketing, customer interaction, claims processing, fraud detection, and underwriting. One use case for the application of AI lies in marketing and customer acquisition.

15 Applications of Artificial Intelligence in Marketing - Smart Insights Digital Marketing Advice


Artificial intelligence means any technology that seeks to mimic human intelligence, which covers a huge range of capabilities such as voice and image recognition, machine learning techniques and semantic search. All the techniques are'AI' in the sense that they involve computer intelligence, but we've broken them down into 3 different types of technology - Machine Learning Techniques, Applied Propensity Models, and AI Applications. A brand that nails voice search can leverage big gains in organic traffic with high purchase intent thanks to increased voice search traffic due to AI driven virtual personal assistants. Programmatic Media buying can use propensity models generated by machine learning algorithms to more effectively target ads at the most relevant customers.

Is this metastasis? – Peyton Rose – Medium


These algorithms usually require enormously large training datasets to achieve accurate classifications, but once trained, they can be efficiently repurposed for different applications using a technique called transfer learning. The data I am working with are composed of 6732 individual tissue slide frames, equally split between tissue classified as normal and tissue classified as metastatic. After visually inspecting both the normal and the metastatic tissue images in my training set, I engineered the following computer vision features: nuclei density, tissue discontinuity, and color compactness. Of the 71 misclassified slides, 40 are metastatic tissue misclassified as normal and 31 are normal tissue misclassified as metastatic.

The steps in the machine learning workflow


Each problem is unique, so it can be challenging to manage raw data, identify the right data to include in the model, train multiple types of models, and perform model assessments. Machine learning uses algorithms that learn from data to help make better decisions; however,it is not always obvious what the best machine learning algorithm is going to be for a particular problem. Examples of machine learning techniques include clustering, where objects are grouped into bins with similar traits; regression, where relationships among variables are estimated; and classification, where a trained model is used to predict a categorical response. Figure 1: Examples of machine learning include clustering, where objects are grouped into bins with similar traits, and regression, where relationships among variables are estimated.