Bus and rail transportation providing company Baselland Transport AG uses Nokia’s computer vision and artificial intelligence threat monitoring system to secure rails and railway crossings in Switzerland.
The companies claim that the deployment of an AI-powered monitoring system to secure rails is the first of its kind in Europe. According to a report, more than 200 loss of lives and over 300 serious injuries had been recorded across several European countries in 2018.
Nokia’s system gathers information in real-time using various technologies to generate insightful analytics that helps in providing better security and safety for passengers and vehicles as it remains a massive concern for authorities due to high chances of deaths and injuries.
Head of rail at Nokia, Karsten Oberle, said, “This project enabled us to address many of the level crossing safety issues which are at the top of priority lists for rail operators. By integrating machine learning into level crossing systems, we will be able to continuously improve and refine safety processes in real-time.”
The artificial intelligence-powered detection system analyzes footage collected from CCTVs to help authorities better understand which activities are normal or abnormal. The system automatically sends notifications to desired authorities regarding any threats in railway crossing regions.
In addition, Nokia’s threat detection tool also recognizes object types, stores event-based video clips, images, and related data. Currently, the system has been deployed in the Münchenstein municipality located in the Arlesheim district of Switzerland.
Head of maintenance electrical systems of Baselland Transport, Michael Theiler, said, “Level crossings are notoriously difficult areas to ensure the safety of passengers, pedestrians, train operators, and motorists.” He further mentioned that Nokia Scene Analytics acts as an intelligent set of eyes to provide critical information in real-time for the purpose of accident prevention.
Data privacy has become a serious topic of discussion at AI Conferences, board meetings of tech companies, and government events. While it seems like an elusive thing to achieve, researchers have proposed a number of technologies that can help us get close to this goal. One of the promising solutions is federated learning, where data computations are moved closer to data source than cloud.
Federated learning is a learning technique that allows users to collectively reap the benefits of shared models trained from rich data without the need to centrally store it, according to Google’s research paper Communication-Efficient Learning of Deep Networks from Decentralized Data. As stated above, federated learning brings the model to the data source (generally edge nodes) rather than bringing the data to the model. It then connects multiple computational devices into a decentralized system, allowing individual data collection devices to assist model training. This allows devices to collaboratively learn a shared prediction model while keeping all training data on the individual device. In this process, federated learning eliminates the need for large amounts of data to be moved to a central server for training purposes. As a result, it addresses our concerns about data privacy.
It is evident that artificial intelligence and machine learning models are data-hungry. And to build efficient and personalized AI-powered solutions, we need access to qualitative and quantitative data. Unfortunately, most of the data required by organizations lie scattered across different companies, user devices, nations, and more. Though, integrating all data sources sounds like an alluring easy fix, it is not practical due to constraints like security, user privacy, governmental regulations, and more. Federated learning seems like a lucrative alternative to these woes while complying with privacy regulation laws. Since data never leaves its original location, federated learning allows diverse data owners to interact and share their data at the organizational level. Further, federated learning has the potential to empower organizations to become less reliant on individual data monopolies while simultaneously generating more money from their own data via ‘protected sharing.’
There are two distinct categories of federated learning.
In instances when data sources share the same feature space but differ in samples, horizontal federated learning is used. Two regional banks, for example, may have quite diverse user groups from their separate regions, with a relatively tiny intersection set of users. However, because their businesses are so similar, the feature spaces are identical.
On the other hand, when two data sets share the same sample ID space but differ in feature space, vertical federated learning can be used. For instance, there are two separate companies in the same city: one is a bank, and the other is an e-commerce business in the same city. Because both user sets are likely to include the majority of the area’s people, their userspace intersects widely. However, because the bank keeps track of the user’s expenditure patterns as well as their credit score, and the e-commerce store keeps track of the user’s browsing and purchase history, their feature spaces are distinctive. To better understand vertical federated learning, consider Client A (Zomato) has information about the customer’s food item purchases on Zomato, while Client B (Dineout) has information about the customer’s restaurant reviews in Surat. Combining these two sets of datasets from different domains can better serve customers by using restaurant reviews information (Dineout) to provide better food recommendations to customers browsing food items on Zomato.
We also have federated transfer learning, whose architecture is an extension of vertical federated learning. Here, in the feature and sample dimensions, the data of various individuals do not overlap very substantially. To put it another way, it implies that datasets on separate clients have diverse sample spaces as well as feature spaces. Federated transfer learning may be used to train a custom model, such as movie recommendations, based on the user’s previous web browsing activity.
In both scenarios, the data owners can interact without jeopardizing the privacy of their individual clients. Vertical federated learning, in particular, offers a great application promise for partnering enterprises that possess data from the same set of users but have disjoint features to train models without disclosing their private data jointly. This allows businesses to build a cluster of multiple servers to participate in the federation with the increasing volume of training data and model size. Unlike horizontal federated learning, vertical federated learning enables a better and deeper understanding of the data field.
However, having too many network transfers inside each organization’s cluster might considerably influence the overall performance of the vertical federated learning operation.
Since vertical federated learning involves identifying the shared data items shared by all parties to prepare the training data, it relies on Private Set Intersection (PSI). PSI detects the intersection of training samples from all parties by using personally identifiable information (e.g., email) as sample IDs to align data instances. As a result, PSI makes intersection sample IDs public to all parties, allowing each party to verify if the data entities displayed in the intersection are also visible in the other parties, i.e. intersection membership. However, industries like healthcare may not be in favor of making information about their user base public. To solve this, researchers presented a framework based on Private Set Union (PSU) in a study published in 2021 that allows each participant to retain sensitive membership information to themselves. Rather than determining the intersection of all training samples, the PSU approach creates training instances from the union of samples.
Meanwhile, vertical federated learning is reported to be subject to backdoor attacks, which modify data from rogue agents during training, as well as inference-phase attacks, which influence test data. However, unlike traditional horizontal federated learning, boosting the resilience of vertical federated learning is difficult due to the lack of clear redundancy among the agents. Recently, scientists have come up with an advanced solution, i.e. robust vertical federated learning (RVFL). As per their under-review paper at ICLR 2022. RVFR can restore the underlying uncorrupted features with verifiable assurances under specific situations, therefore decontaminating the model against a wide range of backdoor attacks. In addition, RVFR protects against adversarial inference phase attacks and missing feature attacks.
There are other major challenges that need to be tackled before going for mainstream adoption of vertical federated learning. At present, vertical federated learning algorithms require intricate training procedures and time-consuming cryptographic operations to maintain privacy, resulting in a poor training pace of machine learning models at the edge. Besides, there’s a chance that cybercriminals can fish out private data from gradients of machine learning parameters during vertical federated learning. A catastrophic data breach would occur due to this. While homomorphic encryption is currently at play to prevent this, much development needs to be done.
Oorwin is an AI-powered recruiting platform that seeks to enhance hiring quality, increase recruiter efficiency, and boost revenue by transforming traditional recruitment procedures. It offers a fully-integrated ATS, CRM, and HRMS platform to deliver efficiency and growth for staffing and consulting firms. Recently, we had an opportunity to get in touch with Sai Vignan Malyala, Principal Data Scientist at Oorwin, to understand the company’s Data Science Culture.
5 Ways Data Science is Leveraged at Oorwin
To improve user efficiency: Oorwin strives to build features that improve the efficiency of the recruiter/user and close more in less time accurately. In this context, some of its features are resume parsers, job description parsers, other document parsers, recruitment pre-screening chatbots, email classifiers with insights generation, personality insights, job closure probability, salary prediction, etc. All of these are important and make users’ work pretty easy and save their time.
Recommendation engines to improve users’ choices: Typically, handling a lot of data and making decisions, especially on jobs, can seem daunting for users. For instance, an individual picks out favorite songs and curates a workout playlist. For its business clients, Oorwin has built recommendation systems like semantic search, similar profiles, and similar jobs.
To grab data and give insights: Oorwin has built more than 60 impeccable scraping engines working all day gleaning different market data for better analysis and insights. This data is used in its model training and the rest to provide market insights necessarily via modeling. Some of these data analysis services include Job Grabber & Insights, Profile Grabber & Insights. At present, Oorwin is working on a recruitment analytics solution that will help users make the right decisions at the right time as per their data and business actions.
Business Analytics to make data-driven decisions: Companies may build plenty of data science use cases with a lot of effort, but it would be worth only if users benefit. That’s why most of the companies have only a 2% success rate on their features. Sai Vignan believes the ‘wow factor is not the key, but the need of the feature.’ Consequently, Oorwin always ensures to analyze its feature’s usage and trends. Sai reveals, “We have a good cluster set for this. Fortunately, we really get great trends insightful enough to choose our next feature to build or enhance. We always have a team doing continuous POC’s checking the need and possibility of different use-cases.”
Data Pipelines & Engineering: The data science team at Oorwin is well equipped with Data Engineering skills to set up data pipelines for training models and deploy those models in production using variant methods as per model.
Fostering Innovation with Disruptive Technologies
As consumer behavior has changed since last year, companies had to introduce new additions in their tech pipeline to cater to the new normal of client expectations. Since last year Oorwin has taken drastic measures to boost collaboration with its clients. For instance, the product usage trends and analysis showcased in dashboards have helped Oorwin identify better applications and use-cases of their solutions during the pandemic. Employees at Oorwin tried to understand what was needed in the product for users during the pandemic. They observed that healthcare-related recruitment had grown more than 40% during the pandemic. This observation inspired the company to improve its models in that domain. Some of the tools used by Oorwin include Python, Streamlit, Power BI, PyTorch, spaCy, hugging face transformers, Gensim, FastAPI, AWS for Dockerization, AWS Lambda, AWS Metrics, and more. Besides, they use custom tools for tagging, training and monitoring data pipelines.
Cultivating the Right Mix of Data Science skills
“To become a successful data scientist, one has to become a good data analyst first. Unless one understands what data he has and what data he wants as per domain knowledge, he cannot build anything meaningful,” exclaims Sai Vignan.
From a generic viewpoint, analytical mindset, ability to build the problem statement, eagerness to learn the data with the right attitude, strong communication skills (to integrate or collaborate with other teams on the use cases) are the primary skills for any successful data scientist.
From a technical viewpoint strong foundation in statistics, algorithms, programming, and domain expertise is a must for successful data scientists.
According to Sai Vignan, there isn’t any ready-made approach to build something in data science. It’s all about the ability to draw insights and create something new from the data. The work of a data scientist comprises both tedious data cleaning and interesting model-building. Therefore, the right mindset is required to excel in this profession.
Data Science Hiring Process of Oorwin
Oorwin ideally looks for people with a strong foundation in programming, statistics, and basic ML/DL/NLP algorithms. However, getting the right candidate is always a concern; thus, the company has a standardized approach of pre-screening and interviewing. In other words, Oorwin has unique criteria of post-selection screening for the right attitude in the candidate suitable for the company. Recruiters favor candidates possessing a blend of skills like humility, dedication, eagerness to learn, and a collaborative demeanor.
Oorwin also boasts of having candidates who are freshers in its workforce. Sai Vignan explains that the eagerness to learn and grow with humility in freshers is unparalleled compared to candidates having industry experience. Oorwin still follows that culture of hiring some freshers, training them internally or externally, and offering them opportunities to develop impressive solutions for the company. Though initially, it will be pretty challenging to build accurate models. Sai Vignan advises data scientists not to get discouraged but to keep the determination to learn from mistakes and make better models in the future. The company also values data science certification during hiring. However, evaluation is purely based on skills and what candidates learned in certification courses.
Sai Vignan advocates that data science for any product should not be a ‘wow’ factor but a need factor. Building anything attractive but not useful for the users will last only for a few days. Focus on building data science solutions that have long-term usability. Data science use-cases take a good amount of time to build. Companies do give the time needed but, if the data science leaders do not use it efficiently to build the right thing, time will be at stake for the product to continue in the market.
Therefore, he encourages building something that eases the user’s work and not that pleases the developer. He also suggests budding data scientists make decisions about their respective work/projects based on their statistical analysis instead of relying on a HIPPO (highest paid person’s opinion).
Data analytics firm Course5 Intelligence files draft papers to raise around Rs 600 crore through its initial public offering (IPO). The company has filed the preliminary documents with the Securities and Exchange Board of India for its IPO.
The draft red herring prospectus mentions that Course 5 Intelligence has planned to provide fresh equity shares of Rs 300 crores and an offer for sale of up to Rs 300 crores.
The company announced that Ashwin Ramesh Mittal, Riddhymic Technologies, Riddhymic Technoserve LLP, AM Family Private Trust, and shareholder Kumar Kantilal Mehta would sell their shares.
The proceeds of the new issuance will be used to fund inorganic growth plans, working capital requirements, product and IP activities, global development, and general corporate reasons, according to the company.
Mumbai-based data analytics company Course5 Intelligence was founded in 2000. It has a vast customer base, including numerous industry-leading companies like Lenovo, Colgate-Palmolive Company, American Regent, Inc (a member of the Daiichi Sanyo Group), and National Bank of Fujairah PJSC.
The company helps businesses in their digital transformation process using various technologies, like artificial intelligence, analytics, and more. Course5 Intelligence is a digital, marketing, and customer analytics firm that specializes in understanding the omnichannel customer journey.
Indian transportation vehicles manufacturer Ashok Leyland announces that it has partnered with autonomous mobility technology developing firm Aidrivers to accelerate the development of artificial intelligence-powered autonomous vehicles.
The companies have signed a memorandum of understanding to jointly develop high-tech self-driving cars as the industry is witnessing an all-time boom. According to the contact, Aidrivers will integrate its autonomous mobility technology with Ashok Leyland’s transportation vehicles to convert them into self-driving vehicles.
The partnership will considerably boost the development of industrial mobility equipment and other autonomous industrial automation solutions for various industries, including logistics. President and CTO of Ashok Leyland, Dr. N. Saravanan, said, “Ashok Leyland has a well-established record as a pioneer in the commercial vehicle sector, having developed many concepts that have become industry benchmarks and norms over the years. For example, we launched India’s first electric bus and India’s first Euro 6 compliant truck.”
He further added that they are incredibly proud to partner with Aidrivers to develop innovative autonomous driving solutions that will drive the industry forward in the years to come. The contract also mentions that the two companies will also join hands for combined marketing efforts.
United Kingdom-based artificial intelligence-powered autonomous mobility technology developing company Aidrivers was founded by Rafiq Swash in 2018. Aidrivers aims to develop autonomous mobility technology to meet industry needs for optimization, resiliency, and safety.
“There are clear opportunities where the skills, resources, reputation and technical knowhow of both Aidrivers and Ashok Leyland can complement each other, particularly in meeting the needs of industry for a sustainable future. Across all sectors, industry is seeking innovative ways to deliver the best efficiency, minimise carbon and environmental impact and reduce costs,” said founder and CEO of Aidrivers Rafiq Swash. He also mentioned that this new partnership with Ashok Leyland lays the foundation of joint innovation and inspiration.
The United Kingdom announces the launch of its new initiative to develop a global standard for artificial intelligence. The United Kingdom’s government has chosen The Alan Turing Institute, supported by the British Standards Institution (BSI) and the National Physical Laboratory (NPL), to carry out this artificial intelligence initiative.
Along with the two institutions, the artificial intelligence hub will also be backed by the Department for Digital, Culture, Media, and Sport (DCMS) and the Office for AI (OAI). As a part of the country’s national artificial intelligence policy, the selected organization will develop an AI hub to boost the contribution of the UK in setting global technical standards for artificial intelligence technologies.
According to the government, new research pointed out that over 1.3 million UK businesses will use AI by 2040, and a massive rise in expenditure in artificial intelligence can be expected during the same period. The newly launched initiative will play a crucial role in facilitating innovations in AI technology and also tap the economic potential of the rapidly growing industry.
DCMS Minister for Tech and the Digital Economy, Chris Philp, said, “It marks the first step in delivering our new National AI Strategy and will develop the tools needed so organisations and consumers can benefit from all the opportunities of AI. We want the UK to lead the world in developing AI standards.”
He further added that the country must remain at the forefront of this AI transformation which is already improving ease of life and can create innumerable job opportunities in the UK.
A prime motive of setting AI standards is to ensure the safe deployment of artificial intelligence solutions in the country, as there are several concerts related to the safety and ethical use of AI solutions. Last year, the UK government also published the world’s first roadmap to accelerate the growth of the artificial intelligence ecosystem in the country.
“The transformative impact of AI is quickly becoming central to our economy and society, already playing a key role in everything from climate science and medical diagnostics to factory robotics and climate change mapping,” said BEIS Minister for Science, Research and Innovation, George Freeman.
Indian data science firm TheMathCompany raises $50 million in its recent funding round led by Brighton Park Capital, an investment firm that primarily funds growth-stage software, healthcare, and tech-enabled services providing companies. Other investors, including Arihant Patni, also participated in the latest funding round.
TheMathCompany plans to use the fresh funds to expand into other global markets like Europe and the United States. The firm will also utilize the funds to further enhance its platform named Co.dx.
The platform is a highly capable solution that drives value for business through analytics at speed and scale. Avendus Capital and Shardul Amarchand Mangaldas & Company served as financial advisor and legal counsel to TheMathCompany during the funding round, respectively.
CEO of TheMathCompany, Sayandeb Banerjee, said, “We are thrilled to announce Brighton Park Capital’s investment. The firm’s deep technology sector expertise and proven track record of scaling global enterprises will be instrumental to our growth.”
The Bengaluru-based technology company The MathCompany was founded by industry veterans Aditya Kumbakonam, Anuj Krishna, and Sayandeb Banerjee in 2016. The firm has been recognized several times as one of the fastest-growing analytics companies in the world. It specializes in helping companies buck age-old trends and create demonstrated value with the help of robust analytics transformations.
TheMathCompany hires talented data scientists and engineers across the globe from leading institutions and provides them with best-in-class training once they get on board. TheMathCOmpany works with Fortune 500 companies operating in multiple industries, including automotive, CPG, retail, technology, manufacturing, and many more. “TheMathCompany is positioned for continued success as it strengthens its brand recognition and expands in the US and EU markets, supporting customers on their missions to capture value through data analytics,” said Managing Partner of Brighton Park Capital Mark F. Dzialga. He also mentioned that they look forward to partnering with TheMathCompany and its team to support its growth.
Technology giant IBM announces that it has acquired data and analytics software developing firm Envizi to help organizations across the globe accelerate their sustainability initiatives and achieve environmental goals.
However, the companies did not provide any information regarding the valuation of this deal. IBM plans to integrate Envizi’s platform with its own to offer unmatched features to its customers.
The acquisition will allow IBM to further enhance the capabilities of its IBM Maximo asset management solutions, IBM Sterling supply chain solutions, and IBM Environmental Intelligence Suite to help organizations build sustainable supply chains.
General Manager of IBM AI Applications, Kareen Yusuf, said, “Envizi’s software provides companies with a single source of truth for analyzing and understanding emissions data across the full landscape of their business operations and dramatically accelerates IBM’s growing arsenal of AI technologies for helping businesses create more sustainable operations and supply chains.”
He further added that it is crucial for organizations to transform data into predictive insights that help them make more intelligent and actionable decisions. The integration of Encizi with IBM’s AI software suites will now allow companies to automate the feedback generation process between their corporate environmental initiatives and the operational endpoints, allowing organizations to accelerate their sustainability efforts.
United States-based data and analytics solution provider Envizi was founded by Andrew Lamble, Bill Clasquin, and David Solsky in 2004. The company specializes in providing software solutions that help businesses optimize resources across several operations. The firm has more than 160 clients that operate in over 160,000 locations worldwide in 112 countries.
Envizi offers a fully customizable and user-friendly dashboard that helps customers effectively manage and analyze their environmental goals. The software can collect over 500 types of data and supports major sustainability reporting frameworks. “As part of IBM, we feel more confident than ever that we can achieve our goal of providing clients and partners with the world class tools they need to reduce their operational impacts and optimize for the low carbon future,” said the CEO and Co-founder of Envizi David Solsky.
Artificial intelligence solutions developing company AI Clearing announces the launch of its new AI Surveyor platform purpose-built for large-scale solar infrastructure projects.
The new initiative of the US government for utilizing non-carbon emitting sources of energy to meet all of the country’s power demand by 2035 can considerably increase the adoption of solar infrastructures in the country.
The newly launched AI Surveyor platform of AI Cleaning is capable of providing intelligent business reports every day based on predictive analytics. The technology uses various sources like drones, GIS, and design information to collect and analyze data for providing daily updates to users regarding the productivity and progress of their solar infrastructure.
Co-founder and Chief Technology Officer of AI Clearing, Adam Wisniewski, said, “Our AI model is trained on data from across the globe, so whether it is the desert of Abu Dhabi or snow in Canada, it can track progress across your KPI’s with an accuracy of 99.98%.”
He further added that as the solar energy market is rapidly growing, numerous interested individuals and companies reach out to them daily with the intent to use AI Clearing’s artificial intelligence-powered solution during the construction of utility-scale solar farms. To date, AI Surveyor has provided more than 300 progress reports and has tracked the completion of over two million solar panels.
United States-based automated analytics for large-scale construction and infrastructure projects providing company AI Clearing was founded by Adam Wisniewski, Michael Mazur, and Michael Mazur recently in 2019. Since its launch, the company has raised $2.5 million from investors like Tera Ventures, Inovo Venture Partners, and Innovation Nest over three funding rounds.
“While production tracking is important for any project, solar construction requires much closer attention to production rates. AI Clearing has proven to be a great partner in this space,” said senior manager of business technologies at PCL, Alex Ramirez.
He also mentioned that solar power plants are gaining immense popularity in the United States as the country is looking for cleaner energy sources.
Microprocessor for AI and ML applications developing firm Graphcore opens its first office in India while the country is witnessing an artificial intelligence revolution. The new office will operate under the leadership of Sudhakar Yerneni. In 2020, Graphcore raised $222 million in its series E funding round led by Ontario Teachers’ Pensions Plan Board, Fidelity International, and Schroders.
All India’s major institutions and enterprises are rapidly promoting education and the adoption of artificial intelligence solutions, making India one of the global hubs for artificial intelligence, machine learning, and data science.
With Graphcore’s office in India, companies will have access to leading-edge ai compute services, which would further drive the growth of artificial intelligence in the country. Sudhakar Yerneni said, “India is primed to take advantage of the AI revolution. This technology has the power to transform our economy and ensure our continued status as a leading research nation in sectors such as healthcare, finance, energy, agriculture and beyond.”
He further added that Graphcore’s computing systems are boosting the pace of artificial intelligence innovations across the globe as the company’s Intelligent Processing Unit users are making breakthroughs, which are not possible on other technologies.
The company looks forward to seeing the kind of innovations its Indian customers will come up with using Graphcore’s technology. The newly opened office of Graphcore is located in Pune, Maharashtra, and is currently curating talented sales and technical support workers to accelerate the growth of Graphcore’s robust artificial intelligence system adoption in the country.
The Center of Development of Advanced Computing (C-DAC), an autonomous scientific society that operates under the Ministry of Electronics and Information Technology of India, has become one of the newest customers of Graphcore in the country. Senior director and HoD HPC Technologies, C-DAC, Sanjay Wandhekar, said, “C-DAC exists to advance India’s advanced compute capability, so it is only natural that we would want to explore the many possibilities offered by Graphcore’s IPU.” He also mentioned that with the delivery of the IPU-POD system for evaluation, they would be able to make the technology available to various parts of the country.
