Singularity Pulse  
     
  Information Technology

Google to Test Biases in AI Data

Google has released a new tool called the “What-If-Tool” that helps in keeping a check on biases in data meant for AI learning. Conventionally, it is required to write bias checking codes separately for each potential bias. This process is time consuming and complex to understand by non-coders. The What-If-Tool allows comparison of decision-making factors of algorithms with another. This allows to see a clear boundary over which an algorithm will prefer one decision over other.

 
  Image  
 

AI to Improve Auditors’ Productivity and Methodology

Oversea-Chinese Banking Corporation (OCBC) Bank is deploying solutions using AI technique to increase productivity of their auditors. The AI algorithm helps in better detection of trading anomalies through analysis of trade data. Analysis is done using ML algorithms. Bank has partnered with two fintech companies and one of them has developed 100+ market abuse indicators. Flagged data points are fed into the supervised ML system to refine the market abuse parameters. It also uses unsupervised ML to detect both known and unknown types of trade anomalies. Currently, it takes around three months to finish audit of trading activities. With these new AI solutions, bank expects a faster turnaround to integrate data sets and analyze them, and the ability to attain better insights that were previously unknown to the auditor

Banking  
  Image  
  Technology

Neural Network to Be Grown in A Petri Dish

An interdisciplinary team of biologists and computer engineers have been awarded a grant of $ 500000 to develop a computer out of living cells and further program it to perform computational tasks. Researches will make use of “optogenetics” to train the system to recognize handwritten digits. It is a biological technique that uses light to control cell movement. Although there has been resistance from ethicists when scientists developed lab-grown globules of human brain tissues. Ethicists are worried that these “mini-brains” could experience consciousness once they become complex.

 
  Image  
 

ML Algorithm to Discover Electrocatalyst

University Associated: Carnegie Mellon University
Harnessing renewable energy requires converting simple chemicals into valuable compounds which can serve as source of energy. Electrocatalysts are used to speed up this process. Researches at Carnegie Mellon University are using ML for the same purpose. Zack Ulissi, assistant professor of chemical engineering, has developed a system to speed up calculations which were done manually earlier. Speedy calculations provide suggestions for new alloys which can be used as an electrocatalyst. The method was tested on discovery of intermetallics that are good electrocatalysts for CO2 reduction and hydrogen evolution. The ML algorithm quickly analyzes combination of intermetallics for properties associated with good electrocatalysts.

Research  
  Image  
  Environment

AI to Speed Up Climate Calculations

Climate simulations consume a lot of computational power and time. Research at Ludwig Maximilian University of Munich have used AI and presented a new algorithm which model interactions in atmosphere faster and without much loss of reliability. While general models simulate atmosphere on grids with dimensions of 50kms, even the best supercomputers find it too complex to model factors like cloud behaviour and carry out relevant calculations. Researchers have used neural networks, a form of ML to design cloud behaviour. Algorithm was trained on data from high-resolution simulations that also included heat transfer of clouds. Further, is it planned to train it on real data.

 
  Image  
 

AI Powered Grammarly App to Be Available With Google Docs

Grammarly, a grammar check service will now be available with Google Docs through a browser extension. Following the suit of Google, Grammarly depends on AI to suggest corrections. The algorithm is combined together with rules, patterns, and AI techniques namely machine learning, deep learning, and natural language processing to provide and improve its suggestions.

Information Technology  
  Image  
  Entertainment

Humans Get AI as Their Stand-Up Comedy Partners

Researcher at University of Alberta has come up with an algorithm that is capable of performing improvised comedy on stage. The training set contained subtitles from thousands of movies. The algorithm can create lines of dialogue in real time basis. It was tested in front of live audience. While majority of people were able to identify lines recited by the AI, the algorithm was able to fool some of them into believing it is a real human being.

 
  Image  
 

Sound Search to Get Better in Song Recognition

Google is rolling out improvements to its Sound Search, a sound recognizing assistant. They are extending technology of “Now Playing”, a successful always-on music recognizer that could be used even without an internet connection. Sound search contains 1000 times more data than Now Playing. With larger data set, focus is on reducing false matches by improving their ML algorithm. It has quadrupled the number of neural networks involved and has also increased the size of each embedding from 96 to 128. The density of these embeddings have also been doubled. It has also adjusted criteria for songs based on their popularity. Popular songs have a lower matching threshold than less popular ones. It ensures more accurate song determination.

Music  
  Image  
  Research

AI Reads Human Brain to Recreate Faces Seen

University Associated: University of Toronto
Neuroscientists at Canada’s University of Toronto are using AI technology to read human minds. The technology reconstructs images based on brain perception. The data is gathered by using electroencephalography (EEG). The test subjects were connected to EEG brainwave-reading equipment and shown images of faces. Brain activity was analyzed using ML algorithms and digitally recreated those images. This is the first attempt to read human minds via EEG data. Earlier studies were done on fMRI technology. Major difference between the two is that EEG is comparatively portable, cheaper and faster. For future, researchers are hopeful that EEG could be deployed to produce “neural-based reconstruction” of images.

 
  Image  
 
Image
“Quantum Computing, a revolution in technology, has shown great promise towards solving complex computing problems currently outside the capabilities of current computers. Though in its infancy, we at Decimal Point Analytics strongly believe that it is going to grow exponentially in the near future. It holds the potential not only to boost the AI revolution but also transform the way data is synthesized. The articles on Quantum computing will give an insight into the recent developments in this space.”
 
  Image  
  Quantum Computing

Tiny Camera Lens for Transferring Quantum Information

Australian National University (ANU) has recently announced the invention of a tiny camera lens that will allow fast and reliable transfer of information from quantum computers to an optical fiber network. The lens is 100 times thinner than a human hair. It is made up of nano-structures forming a metasurface, which controls light with functionalities that outperforms traditional systems. Metasurface camera lens is highly transparent, thus allowing efficient transmission and detection of information encoded in quantum light.

 
  Image  
 

Researchers Prevents Loss Of Quantum Information

One of the disadvantages of quantum mechanics is occasional disappearance of information. A research group has found methods to fully control the information escaping the qubit. Photons are used as the quantum-mechanical equivalents to classic bits, qubits. Information carried by the qubit leaks as result of “decoherence” and “quantum noise”. Researchers determined the path of information flow and also showed that this flow can be controlled. The experiments showed that once the qubit with its environment have been initialized into the right state, it is possible to arbitrarily control information flow and also protect it from disturbances.

Quantum Computing  
  Image  
 
 
 
     
  Banner  
 
Contact us:

5A, B-Wing, Trade Star Building, J. B. Nagar, Andheri-Kurla Road, Andheri (East), Mumbai - 400 059, Maharashtra, India
W: www.decimalpointanalytics.com  
E: info@decimalpointanalytics.com   T: +91 22300 15200