Researchers from Brunel University London and the University of Leicester have developed a novel artificial intelligence (AI) tool that accurately identifies depressed Twitter users by extracting and analyzing 38 data points from the public profile.
The algorithm considers factors like post contents, posting times, and the user’s social circle to determine whether the Twitterati is depressed.
Developers of this AI algorithm claim that it has a respectable accuracy of nearly 90%. Due to various causes, including social stigma or ignorance of mental state, a vast proportion of prospective depression sufferers around the world do not seek professional care.
This negligence then leads to severe delays in diagnosis and treatment. The new AI algorithm can considerably help change the scenario and might open new ways for future diagnosis.
Director of Brunel’s Institute of Digital Future, and co-author of the study, Abdul Sadka, said, “We tested the algorithm on two large databases and benchmarked our results against other depression detection techniques. In all cases, we’ve managed to outperform existing techniques in terms of their classification accuracy.”
The AI tool filters people with fewer than five tweets and then uses natural language processing to repair misspellings and abbreviations in the remaining profiles. According to the researchers, such technology might potentially detect a user’s depression before they publish something online. This would allow social media platforms to raise alerts that might help early diagnose and treat identified users.
Moreover, the bot can also be used for other purposes, including sentiment analysis, employee screening, criminal investigation, etc.
“The next stage of this research will be to examine its validity in different environments or backgrounds, and more importantly, the technology raised from this investigation may be further developed to other applications, such as e-commerce, recruitment examination or candidacy screening,” said co-author of the study Huiyu Zhou.
The study has been published in IEEE Transactions on Affective Computing.
In January 2021, OpenAI launched DALL-E, a 12-billion parameter version of GPT-3 trained to produce pictures from text descriptions using a dataset of text-image pairings. A portmanteau of the artist “Salvador Dalí” and the robot “WALL-E,” DALL-E’s astounding performance was an instant hit in the AI community, and it also received extensive mainstream media coverage. It can also synthesize items that don’t exist in the actual world by combining diverse concepts. Moreover, the DALL-E model can conduct prompt-based image-to-image translation tasks. Recently, OpenAI unveiled DALL-E 2, an upgrade to its text-to-image generator that incorporates a higher-resolution and lower-latency version of the original system.
DALL-E 2 results for “Teddy bears mixing sparkling chemicals as mad scientists, steampunk.” Source: OpenAI
When DALL-E was first announced, OpenAI stated that it will continue to improve the system while looking at possible risks such as picture generation bias and the spread of false information. It was seeking to overcome these challenges with technical precautions and a new content policy, while simultaneously decreasing its computational load and advancing the model’s core capabilities.
An astronaut riding a horse in a photorealistic style (Variation) Source: OpenAI
DALL-E 2 is based on OpenAI’s CLIP image recognition system, which was created to inspect a given image and summarize its information in a way that people can comprehend. OpenAI iterated on this process to generate “unCLIP,” an inverted version that begins with the description and progresses to an image.
Users can now choose and modify particular portions of existing photographs, as well as add or delete items and their shadows, mash-up two images into a single collage, and create variants of an existing image. Furthermore, the output graphics are 1,024 x 1,024 pixels, rather than the 256 x 256 pixels created by the previous version.
The working of DALL-E 2 is divided into two stages: the first makes a CLIP image embedded with a text caption, and the second generates a picture from it. The results are impressive, and they might have a significant impact on the art and graphic design industries, particularly video game businesses, which hire designers to painstakingly develop worlds and concept concepts.
DALL-E 2 produces images that are several times larger and more detailed than the original. This enhancement is possible due to the transition to a diffusion model, a form of image generation that begins with pure noise and refines the image over time, making it a little more like the image requested until there is no noise left at all.
DALL-E can also generate a smart replacement of a given area in an image. Furthermore, you can provide the system with an example image and it will produce as many versions of it as you like, ranging from extremely near copies to artistic revisions.
OriginalEdit 1Edit 2Edit 3Edit 4Edit 5
Unlike the earlier version, which was open for everyone to play with on the OpenAI website, this new version is now only available for testing by verified partners who are limited in what they may submit or produce with it. They are prohibited from uploading or creating images that are “not G-rated” and “could cause harm,” such as hate symbols, nudity, obscene gestures, or “big conspiracies or events relating to important ongoing geopolitical events.” They must also explain how AI was used to create the images, and they cannot share the images with others via an app or website.
The existing testers are also prohibited from exporting their created works to a third-party platform. However, OpenAI aims to incorporate it into the group’s API toolkit in the future, allowing it to power third-party apps. Meanwhile, if you wish to try DALL-E 2 for yourself, you can sign up for the waitlist on OpenAI’s website.
Global technology giant Amazon announces partnership with Johns Hopkins University (JHU) for establishing a new JHU + Amazon Initiative for Interactive AI (AI2AI).
The jointly established artificial intelligence institute will primarily focus on pioneering AI developments, focusing on machine learning, computer vision, natural language processing, and speech processing.
The Johns Hopkins Whiting School of Engineering’s new JHU + Amazon Initiative for Interactive AI will take advantage of the university’s world-class expertise in interactive AI.
According to the announcement, the project’s inaugural director will be Sanjeev Khudanpur, an associate professor in the Department of Electrical and Computer Engineering.
“Hopkins is already renowned for its pioneering work in these areas of AI, and working with Amazon researchers will accelerate the timetable for the next big strides. I often compare humans and AI to Luke Skywalker and R2D2 in Star Wars: They’re able to accomplish amazing feats in a tiny X-wing fighter because they interact effectively to align their complementary strengths,” said Sanjeev.
He further added that he is thrilled about the potential of the Hopkins AI community banding together under the banner of this endeavor and mapping the future of transformational, interactive AI alongside Amazon researchers.
The funding from Amazon will be used for multiple purposes in this project, such as wading annual fellowship to Ph.D. students, collaborating on research projects led by JHU faculty, and several others.
Earlier the two entities had collaborated with four Johns Hopkins faculty members joining Amazon as part of its Scholars program.
“We value the challenges that they bring us and the life-changing potential of the solutions we will create together, and look forward to strengthening our work together over the coming years,” said Ed Schlesinger, Benjamin T. Rome Dean of the Whiting School of Engineering.
The European Union (EU) has given regulatory approval to the first AI tool capable of full autonomous X-ray analysis. Designed by medical imaging company Oxipit, the autonomous AI imaging suite called ChestLink was developed to operate alongside medical practitioners to help with their clinical workflow. ChestLink can scan X-rays for anomalies and give findings to patients if their X-rays are normal. The tool transmits the X-ray to a radiologist for manual assessment when it detects a possible health issue. If the AI is unsure that a patient’s X-ray could show abnormalities, it sends findings to a radiologist as a preventive step to guarantee that they don’t miss any underlying health risks.
According to informative materials from the Oxipit, most X-rays in primary care are trouble-free, therefore automating the procedure for such scans might reduce radiologists’ workloads.
In the EU, the technology currently has CE Class IIb certification, indicating that it complies with safety regulations. The certification is similar to FDA clearance in the United States, but the measurements are significantly different as a CE mark is easier to get, takes less time, and doesn’t require as much review as an FDA clearance. The FDA examines a device to see if it is safe and effective, and it frequently requests additional information from manufacturers.
ChestLink spent more than a year in numerous pilot tests at various locations evaluating 500,000 genuine X-rays before being certified. It made zero “clinically relevant” errors during these tests. Oxipit is hopeful that ChestLink will also get certified by the FDA for potential use in the United States.
The first clinical deployment of ChestLink is scheduled for early 2023. For the purpose of safety and certainty, Oxipit will begin by assigning ChestLink retrospective analyses, i.e., X-rays that have previously been evaluated by radiologists. ChestLink will be deployed for preliminary analysis under the supervision of Oxipit and medical institution personnel after it passes that real-world test. ChestLink will be moved to autonomous prospective reporting in the last stage of implementation, as planned. Staff will be able to swiftly track the processes of application choices using a real-time analytics page if a radiologist needs to manually evaluate a patient’s X-ray after ChestLink granted them a clean bill of health.
Aircraft manufacturing company Boeing partners with technology giant Microsoft to accelerate its digital transformation journey.
This strategic partnership with Microsoft will allow Boeing to use the Microsoft Cloud and AI capabilities to upgrade its IT infrastructure and mission-critical applications with intelligent new data-driven solutions, allowing for new ways of working, operating, and conducting business.
Boeing was one of the first companies to use the Microsoft Cloud, storing multiple digital aviation apps on Microsoft Azure and leveraging artificial intelligence to improve customer outcomes and streamline operations.
According to the plan, Boeing will leverage Microsoft Cloud and AI capabilities to upgrade essential infrastructure, optimize business processes, and several other tasks.
Chief Information Officer and Senior Vice President of Information Technology & Data Analytics at Boeing, Susan Doniz, said, “Today’s announcement represents a significant investment in Boeing’s digital future. Our strategic partnership with Microsoft will help us realize our cloud strategy by removing infrastructure restraints, properly scaling to unlock innovation, and further strengthening our commitment to sustainable operations.”
She further added that Microsoft’s proven collaboration strategy, trusted cloud technologies, and extensive industry knowledge will assist them in achieving their transformation goals and strengthening Boeing’s digital foundation.
This new development will allow Boeing to extract meaningful and long-term value from its vast database, confirming their shared commitment to lead aerospace innovation in the future. Boeing is a global leader in aircraft designing, manufacturing, and servicing commercial airplanes and defense instruments with a customer base spread across 150 countries globally.
EVP and Chief Commercial Officer of Microsoft, Judson Althoff, said, “Boeing and Microsoft have been working together for more than two decades, and this partnership builds on that history to support Boeing’s digital future by helping it optimize operations and develop digital solutions that will benefit the global aviation industry.”
He also mentioned that by providing flexible, agile, and scalable intelligent and data-driven solutions on a secure and compliant platform, the Microsoft Cloud and its AI capabilities would serve as a significant component of Boeing’s digital aviation strategy.
Language learning models are the latest fad in artificial intelligence technologies. In the realm of AI, we’ve seen some pretty remarkable advancements in the last few months. Last year, Google unveiled Pathways, a new AI architecture that works similarly to the human brain and learns faster than previous models. Before, AI models were only trained for one sense, such as sight or hearing, and not both. On the other hand, Pathways allows Google to interpret text, pictures, and audio in a single AI model. Google Pathways was recently put to the test by the Google Research team, who used it to train the Pathways Language Model (PaLM), a 540-billion-parameter dense decoder-only autoregressive transformer, using 780 billion tokens of high-quality text. In “PaLM: Scaling Language Modeling with Pathways”, the team explains PaLM outperforms state-of-the-art few-shot performance on language interpretation and creation tasks in many instances.
Language models anticipate the next item or token in a text sequence based on the preceding tokens. When such a model is applied iteratively, with the projected output fed back as input, the model is called autoregressive. Many researchers have built large language models based on autoregressive concept models that are founded on the Transformer deep-learning architecture. The transformer architecture made it easier for models to capture context when parsing text. This was a game-changer because since the previous language models like Recurrent Neural Networks (RNNs) sequentially analyzed text, training on a vast text corpus had to be done word by word, sentence by phrase, which took a long time. Moreover, it meant that any kind of long-term context was computationally too costly to maintain. Transformer architecture makes use of key, query, and value parameters to determine which portion of the text is most relevant in a given context. Transformer-based models, such as BERT, also use a process known as Attention, which allows the model to learn which inputs require more Attention than others in specific instances.
PaLM is based on a conventional transformer model architecture, although it only employs a decoder and adds the modifications like SwiGLU Activation, Parallel Layers, Multi-Query Attention, RoPE Embeddings, Shared Input-Output Embeddings, and No Biases and Vocabulary.
SwiGLU activations are used for the multilayer perceptron (MLP) intermediate activations, resulting in considerable quality improvements over typical ReLU, GeLU, or Swish activations; and a “parallel” formulation in each transformer block, rather than the standard serialized formulation, results in around 15 percent quicker large-scale training speeds. At autoregressive decoding time, multi-query Attention keeps costs down, and the use of RoPE embeddings instead of absolute or relative position embeddings allows for superior performance on larger sequence lengths. To boost training stability for big models, the system additionally shares the input and output embedding matrices and employs no biases in the dense kernels or layer norms. Moreover, to accommodate the vast number of languages in the training corpus without over-tokenization, the team adopts a SentencePiece vocabulary with 256k tokens.
Any language learning model is based on the idea of using a massive amount of human-created data to train machine learning algorithms that can help build models that replicate how people communicate. OpenAI’s GPT-3, for example, contains 175 billion parameters and was trained on 570 Gigabytes of text. DeepMind’s Gopher, a 280-billion-parameter autoregressive transformer-based dense language learning model was trained on 10.5 Terabytes of MassiveText. This includes various sources like MassiveWeb (a compilation of web pages) C4 (Common Crawl text), Wikipedia, GitHub, books, and news articles. PaLM was trained on a range of English and multilingual datasets, including high-quality online publications, books, Wikipedia articles, interactions, and GitHub code. The researchers also developed a “lossless” vocabulary that retains all whitespace (which is critical for coding), separates out-of-vocabulary Unicode characters into bytes, and divides numbers into distinct tokens, one for each digit.
Regardless of having only 5 percent code in the pre-training dataset, PaLM performs well on coding and natural language tasks in a single model. Its few-shot learning performance is incredible since it is on par with the fine-tuned Codex 12B despite using 50 times less Python code in training. This observation backs up prior findings that larger models can be more sample efficient than smaller models because they can more effectively transfer learning from multiple programming languages and plain language data.
PaLM’s performance may be enhanced even further by fine-tuning it on a Python-only code dataset called PaLM-Coder. With a compile rate of 82.1 percent, PaLM-Coder 540B beats the previous state-of-the-art record of 71.7 percent on a code repair assignment called DeepFix, where the objective is to fix originally erroneous C programs until they compile successfully. It could also decompose multi-step issues into many sections and answer various elementary school-level arithmetic problems. Aside from the astounding feat, PaLM was designed in part to demonstrate Google’s capacity to harness thousands of AI processors for a single model.
PaLM beat other language models on 28 out of 29 English benchmarks, including TriviaQA, LAMBADA, RACE SuperGLUE, etc., improving few-shot performance on language understanding and generation. These tasks include question-answering tasks (open-domain closed-book variant), cloze and sentence-completion tasks, Winograd-style tasks, in-context reading comprehension tasks, and common-sense reasoning SuperGLUE tasks, and natural inference tasks. Furthermore, PaLM displayed remarkable natural language understanding and generating capabilities on several BIG-bench tasks. For example, the model can distinguish between cause and effect, understand conceptual combinations in certain situations, and even guess the movie from an emoji. Even though just 22 percent of the training corpus is non-English, PaLM performs well on multilingual NLP benchmarks, including translation and English NLP tasks.
Also, PaLM demonstrates breakthrough skills in reasoning problems that need multi-step arithmetic or common-sense reasoning by integrating model size with chain-of-thought prompting. PaLM outperforms the previous top score of 55 percent achieved by fine-tuning the GPT-3 175B model with a training set of 7500 problems and combining it with an external calculator and verifier by solving 58 percent of the problems in GSM8K, a benchmark of thousands of challenging grade school level math questions, using 8-shot prompting. PaLM can even provide clear explanations for instances requiring a complicated combination of multi-step logical reasoning, world knowledge, and deep language comprehension. It can, for example, give high-quality explanations for original jokes that aren’t present on the internet.
PaLM is the first large-scale use of the Pathways system, scaling training to 6144 chips, the largest TPU-based system configuration utilized for training to date. Data parallelism is used at the Pod level to scale the training over two Cloud TPU v4 Pods, while conventional data and model parallelism is used inside each Pod. Most earlier language learning models were either trained on a single TPU v3 Pod (e.g., GLaM, LaMDA), employed pipeline parallelism to scale to 2240 A100 GPUs across GPU clusters (Megatron-Turing NLG), or used multiple TPU v3 Pods (Gopher) with a maximum scale of 4096 TPU v3 chips (Megatron-Turing NLG).
PaLM achieves the maximum training efficiency for the language learning model at this scale, with 57.8 percent hardware FLOPs usage. This is due to a combination of the parallelism technique and a Transformer block reformulation that allows for simultaneous computation of the attention and feedforward layers, resulting in speedups from TPU compiler optimizations.
Advanced AI software and hardware developing company BrainChip announces that it has partnered with semiconductor technology and software automation platform SiFive to deploy artificial intelligence technology at the edge.
The companies say they have combined their technologies to offer semiconductor designers artificial intelligence and machine learning computing at the edge.
With high performance, ultra-low power, and on-chip learning, BrainChip’s Akida is a new advanced neural networking processor architecture that takes AI to the edge that previous technologies can not.
At the same time, SiFive Intelligence solutions merge software and hardware to accelerate AI/ML applications with its highly configurable multi-core, multi-cluster capable design.
According to the companies, the two highly advanced technologies will result in a highly efficient edge AI computing solution. For AI and ML workloads, SiFive Intelligence-based processors provide industry-leading performance and efficiency.
Vice President of Products at SiFive, Chris Jones, said, “Employing Akida, BrainChip’s specialized, differentiated AI engine, with high-performance RISC-V processors such as the SiFive Intelligence Series is a natural choice for companies looking to seamlessly integrate an optimized processor to dedicated ML accelerators that are a must for the demanding requirements of edge AI computing.”
He further added that the partnership with BrainChip is a valuable addition to their ecosystem portfolio. Akida, BrainChip’s first neuromorphic processor, replicates the human brain by analyzing only relevant sensor inputs at the point of capture and processing data with exceptional efficiency and precision while consuming minimum energy.
CMO of BrainChip, Jerome Nadel, said, “We are pleased to partner with SiFive and have the opportunity to have our Akida technology integrated with their market-leading product offerings, creating an efficient combination for edge compute.”
He also mentioned that as the company expands its network of portfolio partners, it wants to ensure that these partnerships are based on complementary technologies, enabling capabilities, and a wide range of contexts so that it may reach the maximum number of potential customers.
Celent has recognized TCS Conversa, a conversational AI platform from Tata Consultancy Services (TCS), as a Technology Standout among Retail Banking Intelligent Virtual Assistant (IVA) platforms.
Celent compared ten such IVA platforms, among which TCS Conversa was named the best. The conversation AI platform achieved this because of its advanced technology and numerous unmatched functionalities.
TCS Conversa is a secure, enterprise-ready, and domain-rich conversational platform that enables businesses to easily implement an intelligent conversational assistant for new and current customer interfaces through chat and voice.
An additional advantage of the platform is that it comes with support for both on-site and cloud models.
Senior Analyst at Celent, Bob Meara, said, “A feature-rich, easily deployable platform that provides diverse support capabilities, interactive channel adapters, and on-premise hosting, we find Conversa to be a leading solution for retail banks.”
He also mentioned that Celent considered various factors, including the platform’s functionality, regional availability, technology and integration capability, and customer feedback before making this decision.
According to the report, clients gave TCS a positive overall rating, praising the conversational design elements for their usefulness and the ease of system maintenance in terms of technology.
TCS says that its conversational AI platform is a strong contender for its TCS BaNCS clients, in which the company had already integrated IVA capabilities as a step toward AI democratization. Celent’s report mentions TCS Conversa’s natural language reasoning capability, no-code-dialog design, and workflow as being out-of-the-box.
Business Group Head of Business, Banking, Financial Services, and Insurance, K Krithivasan, said, “Conversational AI is the future of customer experience, and financial services firms want to unlock its full potential. They want powerful, next-generation bots that can process complex queries with a humanized approach. TCS Conversa, a feature-rich advanced AI platform, helps BFSI enterprises transform operations.”
He further added that this award recognizes their vision, market-leading AI capabilities, and widespread use of sophisticated products like Conversa.
Global automobile manufacturing giants General Motors (GM) and Honda announced that they plan to develop new affordable electric vehicles (EVs) to enter a new market.
The partnership entails producing an EV for the North American, South America, and Chinese markets at a lower cost than Chevrolet’s planned Equinox EV.
According to the companies, their jointly developed EVs will be built on a new global architecture that will use next-generation Ultium battery technology.
GM and Honda aim to start mass manufacturing EVs, particularly compact crossover vehicles, in the world’s largest market with annual volumes of more than 13 million vehicles by 2027. On a worldwide scale, GM and Honda will share their best technology, design, and manufacturing strategies to offer affordable EVs.
GM Chair and CEO Mary Barra said, “This is a key step to deliver on our commitment to achieve carbon neutrality in our global products and operations by 2040 and eliminate tailpipe emissions from light-duty vehicles in the US by 2035.”
She further added that the collaboration would allow them to get more people throughout the world into electric vehicles faster than either company could do on its own.
The companies also discuss future EV battery technology collaboration potential for lowering the cost of electrification, increasing performance, and ensuring future vehicle sustainability. Back in 2013, both companies had partnered to develop a next-generation fuel cell system and hydrogen storage technologies.
GM and Honda also collaborated in 2018 to support GM’s EV battery module development efforts. Moreover, Honda mentioned that it is aiming to reach carbon neutrality on a global basis by 2050.
Senior Managing Executive Officer at Honda, Shinji Aoyama, said, “The progress we have made with GM since we announced the EV battery development collaboration in 2018, followed by co-development of electric vehicles including the Honda Prologue, has demonstrated the win-win relationship that can create new value for our customers.”
Technology giant Microsoft recently released a blog to guide users to detect Spring4Shell vulnerabilities across its cloud services.
According to the company, it is currently detecting a ‘limited volume of exploit attempts’ across its cloud services that are aimed at the critical Spring4Shell remote code execution (RCE) vulnerability. Spring4Shell is a zero-day vulnerability (CVE-2022-22965) that security experts have classified as Critical.
It is also known as a proof-of-concept attack that only affects non-standard Spring Framework configurations, such as when Web Application Archive (WAR) packaging is used instead of Java Archive packaging (JAR).
Microsoft’s guide contains all the steps and methods that can be used to identify and rectify the issue.
“Microsoft regularly monitors attacks against our cloud infrastructure and services to defend them better. Since the Spring Core vulnerability was announced, we have been tracking a low volume of exploit attempts across our cloud services for Spring Cloud and Spring Core vulnerabilities,” mentioned Microsoft in the blog.
Below mentioned are the traits of systems that are most vulnerable to the attack –
Running JDK 9.0 or later.
Spring Framework versions 5.3.0 to 5.3.17, 5.2.0 to 5.2.19, and earlier versions
Apache Tomcat as the Servlet container:
Packaged as a traditional Java web archive (WAR) and deployed in a standalone Tomcat instance; typical Spring Boot deployments using an embedded Servlet container or reactive web server are not impacted.
Tomcat has spring-webmvc or spring-webflux dependencies.
People can use the “$ curl host:port/path?class.module.classLoader.URLs%5B0%5D=0” command to determine the vulnerability of their systems.
Though this command can be used as a predictive tool to check vulnerability, any system that falls within the scope of the impacted systems listed above should still be considered susceptible.
/cdn.vox-cdn.com/uploads/chorus_image/image/70716637/DALL_E_Teddy_bears_mixing_sparkling_chemicals_as_mad_scientists__steampunk.0.png)
