Assistive technologies such as handheld tablets and eye-tracking devices are increasingly helping give voice to individuals with paralysis and speech impediments who otherwise would not be able to communicate. Now, researchers are directly harnessing electrical brain activity to help these individuals.

In a study published Wednesday in the New England Journal of Medicine, researchers at the University of California, San Francisco, describe an approach that combines a brain-computer interface and machine learning models that allowed them to generate text from the electrical brain activity of a patient paralyzed because of a stroke.

Other brain-computer interfaces, which transform brain signals into commands, have used neural activity while individuals attempted handwriting movements to produce letters. In a departure from previous work, the new study taps into the speech production areas of the brain to generate entire words and sentences that show up on a screen.

This may be a more direct and effective way of producing speech and helping patients communicate than using a computer to spell out letters one by one, said David Moses, a UCSF postdoctoral researcher and first author of the paper.Related: 

Virtual vocal tract creates speech from brain signals, a potential aid for ALS and stroke patients

The study was conducted in a single 36-year-old patient with anarthria, a condition that renders people unable to articulate words because they lose control of muscles tied to speech, including in the larynx, lips, and tongue. The anarthria was brought on by a stroke more than 15 years ago that paralyzed the man.

The researchers implanted an array of electrodes in the patient’s brain, in the area that controls the vocal tracts, known as the sensorimotor cortex. They measured the electrical activity in the patient’s brain while he was trying to say a word and used a machine learning algorithm to then match brain signals with specific words. With this code, the scientists prompted the patient with sentences and asked him to read them, as though he were tying to say them out loud. The algorithm interpreted what the patient was trying to say with 75% accuracy.

Although the experiment was only conducted in one patient and only included asking the patient to try to say up to 50 words, the study shows that “the critical neural signals [for speech production] exist and that they can be leveraged for this application,” said Vikash Gilja, an associate professor at the University of California, San Diego, who was not involved in the study.

To Moses and his team, this study represents a proof of concept. “We started with a small vocabulary to prove in principle that this is possible,” he said, and it was. “Moving forward, if someone was trying to get brain surgery to get a device that could help them communicate, they would want to be able to express sentences made up of more than just those 50 words.”

STAT spoke with Moses to learn more about the development of the technology and how it could be applied in the future. This interview has been edited for length and clarity.

What problems were you seeking to address?

It’s kind of easy for us to take speech for granted. We have met people who are unable to speak because of paralysis, and it can be an extremely devastating condition for them to be in. It hadn’t been understood before if the brain signals that normally control the vocal tract can be recorded by an implanted neural device and translated into attempted speech.

Can you describe how the technology works? What information goes into it and how is that analyzed to produce words?

This is in no way mind reading; our system is able to generate words based on the person’s attempts to speak. While he’s trying to say the words that he’s presented [with], we record his brain activity, use machine learning models to detect subtle patterns, and understand how those patterns are associated with words. Then we use those models with a natural language model to decode actual sentences when he is trying to speak.

What’s the importance of including the natural language model?

You could imagine when you’re typing on your phone and it figures, “Oh, this might not be what you want to say,” that can be very helpful. Even with the results that we report, it’s imperfect. It helps to be able to use the language model and the structure of English to improve your predictions.

What was surprising to you about what you learned during this study?

One of the very pleasant surprises was that you were able to see these functional patterns of brain activity that have remained intact for someone who hasn’t spoken in over a decade. As long as someone [can imagine] producing the sounds of what their vocal tract would normally do, it’s possible for us to be able to record that activity and identify these patterns.

How did you feel when you realized that the system was in fact producing the words that the patient was trying to say?

My first thought was “OK, that’s just one sentence. It could have been a fluke.” But then when we saw that it was working sentence after sentence. It was extremely thrilling and rewarding. I know that the participant also felt this way, because you can tell from looking at him that he was getting very excited.

What are the next steps to improving the system?

We need to validate this in more than one person. And we want to know how far this technology can go. Can this, for example, be used to help someone who’s locked in completely — who only has eye movements and cannot move any other muscles. If we show that it can work reliably in people with that level of paralysis, then I think that that’s a strong indicator that this is really a viable approach.

How do you envision this technology being applied in the future?

The ultimate goal really for us is to completely restore speech to someone who’s lost it. That would mean any sound someone wants to make, the system is able to produce that sound for them by synthesizing their voice. You could even restore some personal aspects of the speech, such as intonation, pitch, and accent. It’s going to be a lot of effort and we have a lot of work to do, but I think this is a really strong start.

The post Researchers use machine learning to translate brain signals from a paralyzed patient into text appeared first on Artificial Intelligence.

A UW Allen School team recently published an article in Nature Machine Intelligence finding models predicting COVID-19 diagnosis from X-rays are relying on shortcuts.

Several research groups have developed artificial intelligence (AI) models to diagnose COVID-19 based on  chest radiography, with the intention of increasing COVID-19 testing accessibility.

When UW M.D. and Ph.D. students Alex DeGrave and Joseph Janizek heard about this, they immediately thought something in the models might be amiss. 

Janizek said they had been studying AI models predicting pneumonia from chest X-rays and found many of the models were using shortcuts — or aspects of images unrelated to the actual disease — to make the predictions. DeGrave and Janizek thought the COVID-19 diagnosis models might be doing the same thing.  

“We figured that the combination of the high profile of these new studies coming out, and the likelihood of the data being sort of problematic, made it a really good place to kind of apply what we’ve been looking at,” Janizek said.  

Shortcut learning occurs when AI learns to associate things from the training data that are not meaningfully associated in real life. Janizek and DeGrave found the COVID-19 prediction AI models associated having labels on the bottom of the X-ray image with a COVID-19 diagnosis.

DeGrave, Janizek, and their Ph.D. advisor and associate professor Su-In Lee created and trained deep convolutional neural network AI models to replicate what had been done in published studies. The team found the AI performed well on data from the same hospital system as the training data, but when given data from a different hospital system the accuracy was reduced by half.

This trend was something they had noticed in pneumonia models as well, which suggested AI models might be using shortcuts to make diagnosis predictions, Janizek said.

Traditionally, AI functions like a “black box.” The AI model receives large amounts of data to learn from, then users ask the model to make a prediction about a new piece of data. The AI will give an answer, but users typically have no idea why this answer should be reliable.

The team employed a variety of techniques to open up the black box of COVID-19 diagnosis AI models. DeGrave, Janizek, and Lee used saliency maps, which highlighted regions the AI used to determine COVID-19 diagnosis. They also used generative adversarial networks, which involves illustrating what the AI “thinks” is important about the image, in addition to manually modifying the image to see how AI’s COVID-19 diagnosis would change.

“The reason why we used this large set of techniques, three complimentary techniques, is because I think they all overlap each other’s pitfalls nicely,” DeGrave said. “They really complement each other and make the set of experiments much stronger.”

The team found the AI models were using parts of the image, such as annotations, labels, and body positioning, that had nothing to do with COVID-19 to make a COVID-19 diagnosis prediction. These AI models were particularly reliant on shortcut learning because in the limited data available, X-ray images from COVID-19 positive and COVID-19 negative individuals were from different sources.

“Having a problem this severe is fairly unique to COVID,” DeGrave said. “However, there’s a less severe version of the problem that we just see all over the place as well.”

Applying AI to make diagnosis predictions is a popular area of study, but DeGrave said to “be wary also of any other models for any other conditions that were trained in the problematic nature exposed in this paper.”

Janizek said he was surprised when he discovered people were planning to use these problematic models in a clinical setting.

“There needs to be more of these kinds of watchdog type papers, where people are really looking at the reproducibility of existing models and problems that exist out there,” DeGrave said.

The post UW researchers find COVID-19 diagnoses made by artificial intelligence unreliable appeared first on Artificial Intelligence.

The enormous potential of Big Data has already been demonstrated in areas such as financial services and telecommunications. An international team of researchers led by the IPK Leibniz Institute has now tapped the potential of big data for the first time on a large scale for plant research. To this end, data from three projects were used to increase the predictive accuracy for yield in hybrid varieties of wheat.

“We were able to draw on the largest dataset published to date, which contains information from almost a decade of wheat research and development,” says Prof. Dr. Jochen Reif, Head of the Breeding Research Department at IPK. The results, which could herald a new era for plant breeding, have now been published in the magazine Science Advances.

Finally, data on more than 13,000 genotypes tested in 125,000 yield plots were analysed. For comparison: In a breeding programme, plants are tested in 20,000 yield plots every year. “It was clear to us that we would have to increase the population sizes in order to ultimately develop robust predictive models for yield,” says Prof. Dr. Jochen Reif, “so in this case it was really once: ‘a lot goes a long way'”. The effort was worth it, he said. “We were able to double the predictive accuracy for yield in our study.”

The research team used data from the two previous projects HYWHEAT (funded by the Federal Ministry of Research and Education) and Zuchtwert (funded by the Federal Ministry of Food and Agriculture) as well as from a programme of the seed producer KWS. Basically, the challenge in such studies is to prepare the information to a uniform quality level and thus enable a common analysis. “Since we were responsible for the designs of the experiments from the start, we were able to plan them in such a way that a small proportion of the same genotypes were always tested across the projects, thus enabling an integrated analysis in the first place,” says Prof. Dr. Jochen Reif.

The scientist is firmly convinced that it pays off to use Big Data for plant breeding and research. “We have ultimately worked on the future of all of us”, says the IPK scientist. “We have succeeded in showing the potential of Big Data for breeding yield-stable varieties in times of climate change.”

According to Prof. Dr. Jochen Reif, the current model study has a significance that goes far beyond one crop type and hopefully heralds a cultural change in breeding. “We were able to show the great benefits of Big Data for plant breeding. However, the possibilities for this are only possible through a trusting cooperation of all stakeholders to share data and master the challenges of the future together.”

Ultimately, this is also the entry point for the use of artificial intelligence (AI). “The successful use of AI also stands and falls in plant breeding and research with curated and comprehensive data. Our current study is an important door opener for this path.”

The post Big data: IPK researchers double accuracy in predicting wheat yields appeared first on Artificial Intelligence.

Most NLP researchers prioritise the development of deep learning models over the quality of training data. The relative lack of attention results in training data picking up spurious patterns, social biases, and annotation artefacts. 

Data curation is the organisation and integration of data collected from multiple sources. The process involves authentication, archiving, management, preservation for retrieval, and representation.

Her paper laid down the arguments for and against data curation.

Why data curation is important

In her paper, Rogers gives the following arguments in support of data curation:

Social biases: Written text may contain all kinds of social biases based on race, gender, social status, age, and ability. Models may learn these biases, and when deployed in real-world scenarios, they may propagate and further amplify them. This puts minority groups at a significant disadvantage. It’s imperative to select data taking sociocultural characteristics into account and promote fair representation of all social groups.

Privacy: Using personally identifiable information in training data can give rise to privacy and security concerns. For example, a study showed GPT-2 memorised personal contact information even when it appeared only on a few web pages. “Deciding what should not be remembered is clearly a data curation issue,” writes Rogers.

Security: Universal adversarial triggers force models to output a certain prediction. A recently discovered phenomenon, this effect affects the training data, compromising even the robust models. Data curation can help avoid this attack. 

Evaluation methodology: For NLP tasks, the test sample comes from the same distribution as the training samples. There is a possibility of the samples getting overlapped. Curation is necessary to ensure no overlapping takes place.

Progress towards NLU: With rapid scaling, we often lose track of the data on which a model is trained. Without data curation, the models may suffer from one of the following issues:

• Falling prey to common perturbations. For example, linguistic phenomenons such as negations.
• Learning spurious patterns in the data.
• Struggling to learn rare occurrences.

Arguments against data curation

Many experts believe data must be used in their natural form to give an unvarnished output. While there is no problem with this argument, Rogers said, it needs more elaboration. “In that case, the “natural” distribution may not even be what we want: e.g. if the goal is a question answering system, then the “natural” distribution of questions asked in daily life (with most questions about time and weather) will not be helpful,” wrote Rogers. She further added there is still a lot of research work that needs to be done before developers can study the world as it is.

Some developers feel their data is large enough for their training set to encompass the ‘entire data universe’. Rogers said collecting all data is impossible as it will pose legal, ethical, and practical challenges

Meanwhile, many are in favour of developing algorithmic alternatives to data curation. As per Rogers, this is a good possibility; however, having such solutions, in the current scenario, could be a complementary approach to data curation rather than completely replacing it.

A few experts believe data curation is part of the process and should not become a task big enough to forget the original purpose of developing a model. Even though the current deep learning systems are better, they still need to train within the range of the training data, Rogers said.

“A perfect dataset would provide a strong signal for each phenomenon that should be learned. That’s not how language works, so we may never be able to create something like that,” she said. While it may be difficult to achieve perfect solutions, it is always possible to improve the models.

Curation means making a decision about what to include and what to exclude. This can be a daunting task and requires a lot of interdisciplinary expertise, Rogers said.

Wrapping up

“We do want more robust and linguistically capable models, and we do want models that do not leak sensitive data or propagate harmful stereotypes. Whether those goals would be ultimately achieved by curating large corpora or by more algorithmic solutions, in both cases we need to do a lot more data work,” writes Rogers. To achieve this goal, the developers have to overcome interdisciplinary tensions and promote truly collaborative spaces.

The post Plea To ML Researchers: Give Data Curation A Chance appeared first on Artificial Intelligence.

Researchers from the University of Toronto and LG AI Research have developed an “explainable” artificial intelligence (XAI) algorithm that can help identify and eliminate defects in display screens.

The new algorithm, which outperformed comparable approaches on industry benchmarks, was developed through an ongoing AI research collaboration between LG and U of T that was expanded in 2019 with a focus on AI applications for businesses.

Researchers say the XAI algorithm could potentially be applied in other fields that require a window into how machine learning makes its decisions, including the interpretation of data from medical scans.

XAI is an emerging field that addresses issues with the ‘black box’ approach of machine learning strategies.

In a black-box model, a computer might be given a set of training data in the form of millions of labeled images. By analyzing the data, the algorithm learns to associate certain features of the input (images) with certain outputs (labels). Eventually, it can correctly attach labels to images it has never seen before.

The machine decides for itself which aspects of the image to pay attention to and which to ignore, meaning its designers will never know exactly how it arrives at a result.

But such a “black box” model presents challenges when it’s applied to areas such as health care, law, and insurance.

For example, a [machine learning] model might determine a patient has a 90 percent chance of having a tumor. The consequences of acting on inaccurate or biased information are literally life or death. To fully understand and interpret the model’s prediction, the doctor needs to know how the algorithm arrived at it.In contrast to traditional machine learning, XAI is designed to be a “glass box” approach that makes decision-making transparent. XAI algorithms are run simultaneously with traditional algorithms to audit the validity and the level of their learning performance. The approach also provides opportunities to carry out debugging and find training efficiencies.

The first, known as backpropagation, relies on the underlying AI architecture to quickly calculate how the network’s prediction corresponds to its input. The second, known as a perturbation, sacrifice some speed for accuracy and involves changing data inputs and tracking the corresponding outputs to determine the necessary compensation.

There is a lot of potential in SISE for widespread application. The problem and intent of the particular scenario will always require adjustments to the algorithm—but these heat maps or ‘explanation maps’ could be more easily interpreted by, for example, a medical professional.

LG’s goal in partnering with the University of Toronto is to become a world leader in AI innovation. This first achievement in XAI speaks to our company’s ongoing efforts to make contributions in multiple areas, such as the functionality of LG products, innovation of manufacturing, management of supply chain, the efficiency of material discovery, and others, using AI to enhance customer satisfaction.

When both sets of researchers come to the table with their respective points of view, it can often accelerate problem-solving. It is invaluable for graduate students to be exposed to this process.

While it was a challenge for the team to meet the aggressive accuracy and run-time targets within the year-long project—all while juggling Toronto/Seoul time zones and working under COVID-19 constraints—Sudhakar says the opportunity to generate a practical solution for a world-renowned manufacturer was well worth the effort.

The post This New Algorithm can Explain Artificial Intelligence (XAI) appeared first on Artificial Intelligence.

Researchers have used advanced AI and large sets of genomic data to unveil how humans have adapted to recent diseases.

The method could also be applied to new pathogens such as the coronavirus that causes COVID-19, helping identify which gene mutations may be associated with more severe cases of the disease.

The study, by researchers from Imperial College London, the Middle East Technical University, Turkey, and the Universita degli Studi di Bari Aldo Moro, Italy, is published today in a Special Issue of Molecular Ecology Resources on ‘Machine Learning techniques in Evolution and Ecology’.

Natural selection is the process by which beneficial gene mutations are preserved from generation to generation, until they become dominant in our genomes – the catalogue of all our genes. One thing that can drive natural selection is protection against pathogens.

However, if a population of people moves from one environment to another, or changes its way of life, gene mutations that are protective against one pathogen could make people susceptible to new diseases.

One example of such a new disease is Familial Mediterranean Fever (FMF), an inherited autoimmune disease that has emerged over the past 20,000 years. FMF is prevalent in southern Europe, the Middle East and northern Africa, where around 50 percent of the people in the region today carry a gene mutation that makes them more susceptible to the disease.

Spotting selection

This prevalence of a seemingly detrimental gene mutation could be the result of two different types of natural selection. One option is ‘incomplete sweep’, where the gene mutation for susceptibility is in the process of being removed from the population, but has not yet been completely eradicated. In this case, natural selection is ongoing.

The other option is ‘balancing selection’, where some potentially detrimental gene mutations for one condition are preserved in the population because they confer some protection against a different disease. In this case, the gene for FMF susceptibility has been associated with protection against the bacteria Yersinia pestis, which causes the plague.

To determine which version of natural selection is at play in FMF, the researchers turned to advanced AI, which is particularly good at spotting patterns or recognising images. They trained their algorithm on datasets that have known values to test its ability to spot patterns.

The team then ran their algorithm on the database for the 1000 genomes project, which holds genomic data for 2,504 individuals from 26 populations, including the relevant ones around the Mediterranean. They discovered that the FMF gene mutations are still prevalent as a result of ongoing selection; they haven’t reached an equilibrium yet and natural selection is still acting.

Old and new diseases

Lead researcher Dr Matteo Fumagalli, from the Department of Life Sciences at Imperial, said: “This is the first tool to test difference between different types of natural selection, finding signals in the genome that have previously been inaccessible.

“Now we have proven that AI can be used to search genomes for subtle patterns of selection, we can use it to further investigate how humans have both adapted to old diseases, like the plague, and relatively new diseases, like FMF.”

One disease area the team are now investigating is the human relationship with coronaviruses. Humans have been living with coronaviruses for at least 50,000 years, and the greater susceptibility some people have to more severe COVID-19 could be a signal of another balancing selection mechanism.

This study was funded by The Leverhulme Trust, Erasmus+, and Imperial College FoNS European Partners award.

The post Artificial intelligence can help spot traces of natural selection appeared first on Artificial Intelligence.

Artificial intelligence (AI) has made it possible for the first time to create fully artificial human genome sequences that are indistinguishable from the DNA of real donors. A European team just created entire sequences of human DNA, using this AI. Their work was published in the journal PLOS Genetics.

An algorithm that can generate artificial human genomes

“Generative neural networks have been used effectively in many different fields over the past decade, including photorealistic imaging,” say the authors of this new work. Applying a similar concept with genetic data, the researchers built their neural networks using the sequences of 2,500 people stored in databases. The system had to generate sequences with similar characteristics and then mix their creations with real ones to see if they could tell the difference. Through training, the artificial genomes created turned out to faithfully reproduce features of the real genomes, such as allele frequencies (the different versions of a gene). One of the biggest challenges of this work was to verify their reliability, said Aurélien Decelle, co-author of this work and a researcher at the University of Paris-Saclay. “So we spent some time studying the statistical properties of the generated sequences,” he explains.

Only sequences, not whole genomes

These “realistic” and “high-quality” genomes are a first, the researchers note in the paper. This type of neural network has already been used in genetics to generate short sequences “on the order of tens or hundreds of base pairs” (the building blocks of our DNA, of which there are about 3 billion in humans), explains Flora Jay, who co-led this work at the University of Paris-Saclay. “But the generation of such long sequences (about 10,000 variants comprising several million base pairs) and in the context of population genetics is new and represents a major step forward,” she adds.

As a result, these artificial genomes “are indistinguishable from the other genomes in the biobank that we used for our algorithm, except for one detail: they do not belong to any real donor,” Luca Pagani, co-author of the study, explains in a press release.

However, the process still needs to be perfected. “One of the main drawbacks is that these models cannot yet be used to create whole artificial genomes due to computational limitations,” and they must be limited to bits and pieces, the authors explain. In addition, very rare alleles are difficult to represent with the algorithm. The final challenge is to “closely monitor the originality of the generated data, i.e., whether they are sufficiently different from the genomes of real donors,” Flora Jay says, adding that this is an ongoing research topic.

Human genome study without concerns for privacy

Far from being without a purpose other than the scientific achievement itself, this type of artificial intelligence can solve the ethical problems associated with genetic databases. “In population genetics, researchers need to regularly compare the data they produce to some reference genomes or sometimes even to a large reference panel. Ideally, these genomes should reflect genetic diversity,” says Flora Jay. Artificial genomes could perform this function reliably and safely.

“Existing genomic databases are an invaluable resource for biomedical research, but they are not publicly available or are protected by lengthy and exhaustive application procedures due to legitimate ethical concerns,” explains author Burak Yelmen. “Artificial genomes can help us overcome this problem within a safe ethical framework.” Looking ahead, Flora Jay predicts that these artificial genomes “will contribute to applications as diverse as understanding our evolutionary past or medical epidemiology by incorporating greater genetic diversity”.

The post French Researchers Develop the First Artificial Intelligence Capable of Creating Human Genomes Sequences appeared first on Artificial Intelligence.

Researchers say that machine learning could help guide healthcare workers’ treatment decisions for lung cancer patients after developing a model that is 71% more accurate at predicting survival expectancy of patients.

A team of Penn State Great Valley researchers conducted a study in which they developed a deep learning model that is more than 71% accurate in predicting survival expectancy of lung cancer patients, which is significantly better than traditional machine learning models that the team tested which have around a 61% accuracy rate.

Deep learning is a type of machine learning that is based on artificial neural networks, which are generally modelled on how the human brain’s own neural network functions.

Informing patient care

The team say that the information on a patient’s survival expectancy could help guide doctors and caregivers in making better decisions on using medicines, allocating resources, and determining the intensity of care for patients. The machine learning model is able to analyse vast amounts of data and can include information such as types of cancer, tumour size, the speed of tumour growth, and demographic data.

Youakim Badr, associate professor of data analytics, said: “This is a high-performance system that is highly accurate and is aimed at helping doctors make these important decisions about providing care to their patients. Of course, this tool can’t be used as a substitute for a doctor in making decisions on lung cancer treatments.”

According to the researchers this deep learning method may be uniquely suited to tackle lung cancer prognosis because the model can provide the robust analysis necessary in cancer research.

Badr said: “Deep learning is a machine-learning algorithm that makes associations between the data, itself, and the labels that we use to describe the data examples. By making these associations, it learns from the data.”

Robin Qiu, professor of information science and engineering and an affiliate of the Institute for Computational and Data Sciences added that deep learning also offers several advantages for many data science tasks, especially when confronted with data sets that have a large number of records, in this case patients, as well as a large number of features.

“In deep learning we can go deeper, which is why they call it that. In traditional machine learning, you have a simple structure of layers of neural networks. In each layer, you have a group of cells,” he said. “In deep learning, there are many layers of these cells that can be architected into a sophisticated structure to perform better feature transformation and extraction, which gives you the ability to further improve the accuracy of any model.”

In the future, the researchers would like to improve the model and test its ability to analyse other types of cancers and medical conditions.

The post Choosing better lung cancer treatments with machine learning appeared first on Artificial Intelligence.

Could giving robots voice help them learn human commands?

Robots have become an integral part of human’s daily lives. They help us in numerous ways, from performing complex tasks to lifting heavy weights and assisting the elderly, playing with kids, and entertaining people at events. They can interact with people in any scenario. However, construing a human language still a challenge for robotic systems. Training them with real-world experiences and knowledge about the world could help robots understand natural language.

People use language to express emotions, direct behavior, ask and answer questions, provide information, and ask for help. Language-based interfaces for robots require minimal user training and expression of a variety of complex tasks.

In a paper, researchers from MIT describes a new way to train machines. They noted that children learn language by observing their environment, listening to the people around them, and understanding what they see and hear. With keeping that in mind, they created a tool called semantic parser that mimics the experience of children learning a language. Parsers are already being used for web searches, natural-language database querying, and voice assistants. The system observes captioned videos and links the words that speakers say with recorded objects and actions.

As parsers are trained in sentences annotated by humans, they could be used to improve natural interaction between humans and robots. According to the paper, a robot equipped with the parser could observe its environment to reinforce its understanding of spoken commands, even when the spoken sentences are not fully grammatical or clear.

Earlier, Analytics Insight reported that how giving voice to robots within healthcare influence human perception. Already, robots are delivering a wide range of healthcare services and opportunities to medical personnel and advancing patient care delivery. In this article, we noted how researchers at the University of Auckland and Singapore University of Technology & Design have been using speech synthesis techniques to create robots that sound more empathetic. As part of their study, researchers tested a hypothesis on how a robot’s voice can impact users’ understanding by conducting a simple experiment using a robot called Healthbot. They used a professional voice artist for the robot’s voice, which was recorded while reading dialogs in two voice variations: a flat monotone and an empathetic voice.

More broadly, teaching a machine to speak and making them able to recognize human voice is a crucial yet effective step as spoken language is the most intuitive form of interaction for humans. In 2018, it was reported that researchers in Japan attempted to bring audition, or power of listening, to robots. Proposed by Tokyo Institute of Technology Professor Kazuhiro Nakadai and Professor Hiroshi G. Okuno of Waseda University in 2000, “Robot Audition” is a research area. For this, they turned their research public and made it open-source software. This essentially helped them generate interest and diversified the research. Their research was officially registered in the IEEE Robotics and Automation Society.

So, when robots and robotics systems are able to learn and recognize the human language, they will have a more emphatic impact on people’s lives.

The post WHY DO ROBOTS NEED TO LEARN LANGUAGE? appeared first on Artificial Intelligence.

Recently a study was published in the journal Nature Chemistry, detailing the outcome of research intended to calculate the ground state of the Schrödinger equation in quantum chemistry. The problem was solved with the application of artificial intelligence techniques, and the success of the study holds major implications for quantum chemistry.

The current method of determining the chemical properties of a molecule relies on slow, resource-intensive, painstaking laboratory experiments. In contrast, quantum chemistry endeavors to predict the physical and chemical properties of molecules, relying only on the arrangement of atoms within a 3D space. In order for quantum chemistry to plausibly determine molecular properties, Schrödinger’s equation needs to be solved. The Schrödinger equation plays the same role that the conservation of energy and Newton’s laws play in classical mechanics, it predicts how a system will behave in the future. Schrödinger’s equation is expressed in terms of a wave function that precisely predicts the probability of an outcome or event. Until now, solving the Schrödinger equation proved immensely difficult.

In order to solve the Schrödinger equation, researchers needed to correctly model a wave function, a mathematical object capable of specifying the behaviors of electrons in a molecule. Wave functions are high-dimensional entities, and as a result, it’s incredibly difficult to encode the relationships between electrons. Some quantum chemistry techniques don’t bother with encoding a wave function, focusing instead on determining the energy of a target molecule. However, an approximation is needed when focusing solely on the energy of a molecule, and this estimation limits how useful predictions can be.

While there are other techniques that quantum chemists can use to represent a wave function, they are essentially true impractical to be useful for calculating the wave function of a few atoms.

According to Phys.org, researchers from Freie Universitat Berlin managed to have solved Schrodinger’s equation with the assistance of deep learning techniques. The research team turned to a “Quantum Monte Carlo” approach, which offers high accuracy at a modest computational cost. The researchers used deep neural networks to represent the wave function for electrons. Professor Franke Noe was the lead researcher on the study, and Noe explained that the neural network was designed to learn the complex patterns regarding how electrons are distributed around the nuclei of an atom.

In order for the researchers to effectively use deep neural networks to learn the patterns behind electrons, they needed to create the right network architecture. Electronic wave functions have a property known as antisymmetry. Whenever two electrons are exchanged the sign of the wave function must change. This particular quirk had to be accounted for and the property baked into the network architecture. The network was named “PauliNet”, getting its name from the “Pauli exclusion principle”. This principle that states that two or more identical fermions can’t exist within the same quantum state at the same time within a quantum system.

PauliNet also had to integrate other physical properties of the electronic wave functions into the network. Instead of allowing the network to come to decision just from observing data, the network had to take the properties of the wave function into account. 

“Building the fundamental physics into the AI is essential for its ability to make meaningful predictions in the field. This is really where scientists can make a substantial contribution to AI, and exactly what my group is focused on.

The research team still needs to conduct more experiments, refining their approach before the model is ready to be applied outside the lab. However, once the method is ready for industrial applications it could be used in a variety of different fields. Materials scientists could use the algorithm to help create new metamaterials, and the pharmaceutical industry could use it to synthesize new kinds of drugs.

The post New Study From AI Researchers Solve Schrodinger Equation appeared first on Artificial Intelligence.