Home Blog

    Corporate America is fretting over Taiwan risks, regulatory filings show


    Executives at publicly traded US companies are becoming increasingly worried about the spectre of a further escalation of tensions over Taiwan, a major supplier of crucial components like semiconductors.

    The number of annual regulatory filings citing Taiwan as a risk factor has risen significantly over the past 12 months, according to Financial Times calculations based on Sentieo data. In March, a popular time for releasing so-called “10-k” reports, 116 companies mentioned Taiwan as a risk to their business, and the rolling 12-month average this month reached its highest level in at least 16 years.

    Technology companies represent the sector most concerned, with those in the semiconductor industry raising the loudest alarm. This is because Taiwan, which is the biggest producer of the most advanced chips, is rapidly becoming one of the world’s most dangerous geopolitical flashpoints. The fear is that in the event of a conflict with China, US firms will be unable to get the microchips needed to make smartphones, electric cars, new weapons, computers industrial machines, and even medical devices. Healthcare is the second most-concerned sector.

    “A ‘de facto’ blockade by Mainland China’s regular military exercises would create bottlenecks in fast-growing sectors dependent on semiconductors, such as high performance computing, internet of things, data centres and electric vehicles,” Alicia García-Herrero, chief Asia-Pacific economist at French bank Natixis, said.

    In a sign of the potentially wide-ranging corporate effects, a clutch of chief executives at big US banks told Congress this week that they would comply with any US government demand to pull out of China if Beijing were to attack Taiwan. The remarks came just days after US president Joe Biden said the US would defend Taiwan from a Chinese attack.

    The median US company had only had five days’ worth of chip inventories in 2021, down from 40 in 2019, according to a study the Department of Commerce.

    At the beginning of August, Biden signed the Chips Act, which will provide $280bn in funding to prop up and kick-start domestic semiconductor manufacturing and research.

    “The US will put more pressure on key suppliers to ban exports to China and develop production in its own market with industrial policy tools, such as the Chips Act and a push for friend-shoring,” García-Herrero said.

    Corporate America is fretting over Taiwan risks, regulatory filings show Republished from Source via

    Top Neural Network Architectures For Machine Learning Researchers


    The neural networks discussed are specifically referred to as artificial neural networks. As the name implies, they are based on what is known about the structure and operation of the human brain.

    A neural network is a computing system composed of several crucial yet intricately linked parts, sometimes called “neurons,” stacked in layers and processing data using dynamic state reactions to outside inputs. In this structure, designs are communicated to one or more hidden layers present in the network by the input layer, which in this structure has one neuron for each component present in the input data. These layers are only referred to as “hidden” because they do not make up the input or output layer. This method is quite helpful in locating patterns that are too complex to manually obtain and train into the computer, as we will see later.


    All of the processing actually takes place in the hidden layers through a network of connections known as weights and biases (W and b): after receiving the input, the neuron calculates a weighted sum while also including the prejudice and then uses the result along with a preset activation function—the most popular of which is the sigmoid function, though better ones like ReLu are also available—to determine whether it should be “fired” or activated. The neuron transmits the data in a “forward pass” process to other linked neurons. At the end of this process, the final hidden layer is connected to the output layer, which has one neuron for each potential desired output.

    Top Neural Network Architectures


    Perceptrons, merely computational representations of a single neuron, are regarded as the initial generation of neural networks. The perceptron, also known as a feed-forward neural network, feeds data from the front to the back. The neuron receives information, processes them, and produces a result. Backpropagation is typically required for perceptron training, providing the network with paired datasets of inputs and outputs. The discrepancy between the input and the outcome is frequently one of the many variations of the error being backpropagated. Theoretically, the network can always model the relationship between the input and output if there are enough hidden neurons. Though practically much more constrained, they are frequently coupled with other networks to create new networks.

    Perceptrons have some limits. However, you can achieve practically anything if you manually select the features and have enough features. We can create any kind of discrimination on binary input vectors since we can have a different feature unit for each of the exponentially many binary vectors. A perceptron’s ability to learn is severely constrained once the hand-coded features have been identified.

    Convolutional Neural Networks

    Contrary to most other networks, convolutional neural networks are rather unique. They can be used for various inputs, such as audio, although their primary purpose is image processing. When you input the network images to classify, that is a typical use case for CNNs. CNNs frequently begin with an input “scanner” that isn’t designed to instantly analyze all of the training data. For instance, you wouldn’t need a layer with 10,000 nodes to input an image with 100 by 100 pixels. Instead, you make a 10 x 10 scanning input layer into which you feed the image’s initial 10 x 10 pixels. After the input has been passed, you move the scanner one pixel to the right to provide the following 10 × 10 pixels.

    Instead of traditional layers, where each node is connected to every other node, convolutional layers are utilized to process the input data. Each node only thinks about cells that are close to it. As they get deeper, these convolutional layers also get smaller, usually due to input components that are easily divided. They frequently include pooling layers in addition to these convolutional layers. A common pooling strategy is max pooling, where we take, for example, 2 x 2 pixels and pass on the pixel with the most red. This method of filtering out details is known as pooling.

    Recurrent Neural Networks

    Perceptrons are essentially what recurrent neural networks (RNNs) are made of; however, unlike perceptrons, which are stateless, RNNs contain connections between passes and connections over time. Because they combine two characteristics—distributed hidden state, which lets them store a lot of historical data quickly, and non-linear dynamics, which enables them to update their hidden state in complex ways—RNNs are incredibly powerful. RNNs are capable of computing anything your computer is capable of computing given enough time and neurons.

    What types of behavior can RNNs display, then? They can oscillate, settle into point attractors, and exhibit chaotic behavior. Additionally, kids might be taught to design a large number of little programs that each capture a piece of information and execute in parallel to interact to create incredibly complex results.

    Long / Short Term Memory

    LSTM networks use gates and an explicitly defined memory cell to try to solve the vanishing/exploding gradient problem. LSTMs have “input gates” that add new information to the cell and “output gates” that determine when to transmit the cell’s vectors onto the following hidden state. Unless a “forget gate” instructs the memory cell to forget those values, the memory cell keeps the initial values and retains them.

    Gated Recurrent Unit

    A minor modification to LSTMs is gated recurrent units (GRUs). Kyunghyun Cho et al. presented gated recurrent units (GRUs) as a gating technique for recurrent neural networks in 2014. The GRU has fewer parameters than an LSTM because it doesn’t have an output gate, but it is similar to an LSTM with a forget gate. It was discovered that GRU and LSTM performed similarly on some polyphonic music modeling, speech signal modeling, and natural language processing tasks. On some smaller, less frequently used datasets, GRUs have been proven to perform better.

    Hopfield Network

    A Hopfield network (HN) is a network in which every neuron is linked to every other neuron; it resembles a spaghetti-like mess since every node serves as every other node. Each node receives input before training, is concealed during exercise, and is output. The weights can then be determined once the networks have been trained by changing the neurons’ values to the desired pattern. Following this, the consequences stay the same. The network will always converge to one of the learned patterns once trained in one or more ways because the network is only stable in those states.

    Boltzmann Machine

    A particular kind of stochastic recurrent neural network is the Boltzmann Machine. It can be thought of as the stochastic version of Hopfield nets’ generative side. It is able to represent and resolve challenging combinatorial problems. It was one of the first neural networks capable of learning internal representations. Like Hopfield Networks, Boltzmann machines have specific neurons labeled as input neurons while leaving others “hidden.” After a full network update, the input neurons change into output neurons. The neurons usually have binary activation patterns when compared to a Hopfield Net. With random weights at first, it learns by backpropagation.

    Deep Belief Network 

    A deep belief network (DBN) is a type of deep neural network used in machine learning. It comprises numerous layers of latent variables, or “hidden units,” with connections between the layers but not between the units within each layer.

    Unsupervised training on a set of instances enables a DBN to develop the ability to probabilistically recreate its inputs. After that, the layers serve as feature detectors. A DBN can be further taught under supervision to perform categorization after this learning phase.


    An artificial neural network called an autoencoder is used to learn effective codings for unlabeled input (unsupervised learning). By teaching the network to disregard irrelevant data (or “noise”), the autoencoder learns a representation (encoding) for a set of data, generally for dimensionality reduction. The encoding is validated and improved by attempting to regenerate the information from the encoding.

    Some variations try to make the learned representations take on beneficial features. Examples include variational autoencoders, which have applications as generative models, and regularized autoencoders (Sparse, Denoising, and Contractive), which are efficient in learning representations for later classification tasks. Autoencoders solve various issues, such as word meaning acquisition, feature detection, anomaly detection, and facial recognition. Additionally, autoencoders are generative models that can generate new data at random, similar to the input data (training data).

    Generative Adversarial Networks

    Generative Adversarial Networks (GANs) are composed of two networks, one of which is charged with producing material (generative), and the other of which is tasked with evaluating content (judgmental) (discriminative). The discriminative model is tasked with deciding if a particular image (one from the dataset) seems natural or artificially produced. The generator’s job is to create images that appear natural and are similar to the initial data distribution. This can be viewed as a two-player zero-sum or minimax game.

    The study uses the analogy that the discriminative model is “the police seeking to find the counterfeit currency.” Still, the generative model is “a team of counterfeiters, trying to make and utilize false currency.” The discriminator tries to avoid being duped by the generator, while the generator tries to trick the discriminator. Both approaches are enhanced due to the models’ training through alternating optimization, where “counterfeits are indistinguishable from the real products.”


    Convolutional neural networks are feed-forward neural network that excels at processing large-scale images because their artificial neurons may respond to a portion of the surrounding cells in the coverage range. LeCun et al. proposed the convolutional neural network topology LeNet in 1998. LeNet is a common term for LeNet-5, a straightforward convolutional neural network.

    LeNet5’s architecture is relatively simple. Image features will be dispersed throughout the whole picture. By combining learnable parameters with convolutions, similar characteristics can be retrieved quite well. LeNet5 was developed when CPUs were extraordinarily sluggish, and no GPU was available to aid in training.

    This architecture’s primary benefit is the reduction of computation and parameter usage. LeNet5 contrasted this with an elaborate multi-layer neural network where each pixel was treated as a separate input. Since the photos have strong spatial correlations, employing single pixels as distinct input features would be problematic and should not be used in the first layer.


    AlexNet is the name of a convolutional neural network (CNN) architecture. It used the non-saturating ReLU activation function, which outperformed tanh and sigmoid in terms of training performance. The initial five layers of AlexNet were convolutional, part of them was followed by max-pooling layers, and the final three layers were fully connected.

    One of the most important studies in computer vision is AlexNet, which inspired many other works using CNNs and GPUs to speed up deep learning. According to Google Scholar, the AlexNet paper has had over 80,000 citations as of 2021.


    Each convolutional layer in Oxford’s VGG networks employed smaller 33 filters for the first time. Also combined as a series of convolutions were smaller 33 filters.

    LeNet’s guiding principles are contrasted by VGG. Large convolutions were used to capture a set of similar features in an image. Compared to LeNet architecture, VGG used smaller filters on the network’s initial layers. Large AlexNet filters like 9 x 9 and 11 x 11 were not applied in VGG. Multiple 3 x 3 convolutions in succession made it possible to mimic the impact of larger receptive fields like 7 × 7 and 5 x 5. It was also VGG’s most important benefit. Multiple 33 convolutions are used in series in modern network architectures like ResNet and Inception.

    GoogLeNet and Inception

    The efficiency of server farm architectures and large deployments have come to the fore for internet behemoths like Google. The ImageNet Large-Scale Visual Recognition Challenge included the GoogLeNet, a 22-layer deep convolutional network, in 2014. GoogleNet is the first architecture developed to lighten deep neural network processing. The content of video frames and images was categorized using deep learning models.


    Please Don't Forget To Join Our ML Subreddit

    <img width="150" height="150" src="×150-1.jpeg" class="avatar avatar-150 photo" alt loading="lazy" srcset="×150-1.jpeg 150w,×80-1.jpeg 80w,×24.jpeg 24w,×48.jpeg 48w,×96-1.jpeg 96w,×300-1.jpeg 300w" sizes="(max-width: 150px) 100vw, 150px" data-attachment-id="17048" data-permalink="" data-orig-file="" data-orig-size="853,1280" data-comments-opened="1" data-image-meta="{"aperture":"0","credit":"","camera":"","caption":"","created_timestamp":"0","copyright":"","focal_length":"0","iso":"0","shutter_speed":"0","title":"","orientation":"0"}" data-image-title="WhatsApp Image 2021-08-01 at 9.57.47 PM" data-image-description data-image-caption="


    ” data-medium-file=”×300.jpeg” data-large-file=”×1024.jpeg”>

    Prathamesh Ingle is a Consulting Content Writer at MarktechPost. He is a Mechanical Engineer and working as a Data Analyst. He is also an AI practitioner and certified Data Scientist with interest in applications of AI. He is enthusiastic about exploring new technologies and advancements with their real life applications

    A new Speech Recognition Pipeline from CMU Research can recognize almost 2000 Languages without Audio


    Voice-to-text processing has advanced significantly in recent years, making the occasional failures in AI-powered speech recognition systems little more than curious outliers. However, most modern speech recognition models depend on sizable supervised training data. Obtaining such statistics is simple for popular languages like English, Chinese, etc. However, it is difficult for the bulk of the 8,000 languages spoken worldwide—low-resource tongues. A research team from Carnegie Mellon University created a voice recognition pipeline that does not need audio for the target language to address this problem. Using 10,000 raw text utterances from the CMU Wilderness dataset, this ASR2K algorithm identifies 1909 languages without audio for the target language and produces remarkable 45 percent CER and 69 percent WER results. The team’s research was also published in the paper, ‘ASR2K: Speech Recognition for Around 2000 Languages Without Audio.’

    The model only assumes that it has access to unprocessed text datasets or a set of n-gram statistics. Three elements make up their speech pipeline: acoustic, pronunciation, and language models. The target languages’ phonemes, including those of unseen languages, are recognized using the acoustic model. In a grapheme-to-phoneme (G2P) model, the pronunciation model forecasts the phoneme pronunciation given a grapheme sequence. Both the acoustic and pronunciation models use multilingual models without supervision, in contrast to the conventional pipeline. In order to apply their newly acquired linguistic skills to low-resource languages without supervision, they can first be trained using supervised datasets from high-resource languages.

    The raw text dataset or n-gram statistics are used to construct the language model. A lexical graph is created by encoding the approximate pronunciation of each word using the pronunciation model. By counting the n-gram statistics, the model can also estimate a traditional n-gram language model thanks to the text dataset. A weighted finite-state transducer (WFST) decoder is subsequently created using this language model in conjunction with the pronunciation model. The team proposed method was then applied to 1909 languages on the Crúbadán: a sizable collection of n-grams for endangered languages.

    The method was evaluated on 129 languages using two separate datasets, Common Voice (34 languages) and CMU Wilderness (95 languages). With Crbadán statistics, it achieved 50% CER and 74% WER on the Wilderness dataset, which were subsequently increased to 45% CER and 69% WER when using 10,000 raw text utterances. The team’s discovery represents a turning point because it represents the first attempt to create an audio-free speech recognition pipeline for tens of thousands of languages. The team’s paper and related code will also be published at the 23rd INTERSPEECH Conference in South Korea.

    This Article is written as a research summary article by Marktechpost Staff based on the research paper 'ASR2K: Speech Recognition for Around 2000 Languages without Audio'. All Credit For This Research Goes To Researchers on This Project. Check out the paper and github link.
    Please Don't Forget To Join Our ML Subreddit

    Khushboo Gupta is a consulting intern at MarktechPost. She is currently pursuing her B.Tech from the Indian Institute of Technology(IIT), Goa. She is passionate about the fields of Machine Learning, Natural Language Processing and Web Development. She enjoys learning more about the technical field by participating in several challenges.

    Treasury Prime Enters Banking-as-a-Service Partnership with FirstBank


    Treasury Prime, a Banking-as-a-Service (BaaS) company, announced it is partnering with FB Financial Corp (FirstBank) in order to bring embedded finance offerings to its customers.

    The partnership with FirstBank brings Treasury Prime’s bank network “to well over a dozen, adding to their rapidly growing available deposit base within the bank network.”

    Chris Dean, co-founder and CEO of Treasury Prime, said:

    “As a modern banking platform, one of our key differentiators, and something that our customers place a lot of value in, is our network of banks, as it provides fintechs the flexibility they want and the scalability they need. By adding FirstBank, we’ve partnered with a financial institution that meets all the key markers to thrive in our network and deliver for enterprises looking for embedded finance.”

    FirstBank is among the first financial institutions “on a Jack Henry core to leverage Treasury Prime’s embedded finance solutions.” The Tennessee-based financial institution “with more than $12.7 billion in assets is expanding its client base by delivering embedded finance to Fintech, Payments and InsureTech.”

    Wade Peery, Chief Innovation Officer of FirstBank, remarked:

    “Treasury Prime’s technology enables us to enhance what we offer our clients and expand our reach to enterprises looking for embedded finance partners. We think embedded banking solutions, like what Treasury Prime offers, are at the forefront of helping the banking industry be more accessible, inclusive and compliant.”

    The FirstBank partnership “comes just months after Treasury Prime announced a similar relationship with Texas-based Third Coast Bank.” The company expects “to make additional announcements related to its bank network in the weeks ahead.”

    As noted in the update, Treasury Prime is “building the future of finance.”

    Leveraging its APIs and versatile Banking as a Service (BaaS) products, Treasury Prime “enables fintechs, banks and corporate partners to innovate, adapt, grow and scale to stay competitive in a rapidly changing marketplace.”

    The company helps fintechs “with a range of complex services including money transfer, risk mitigation and access to a chartered bank’s infrastructure.”

    Treasury Prime works “with forward-thinking banks to extend their boundaries, helping to grow topline revenue and lower the cost of deposits.”

    Treasury Prime Enters Banking-as-a-Service Partnership with FirstBank Republished from Source via

    Crypto Processing Platform Calypso Pay Now Supports Subscriptions, Recurring Payments


    Calypso Pay, an all-in-one crypto processing & acquiring platform, implemented support for recurring payments and subscriptions in crypto.

    Now Calypso Pay customers can “integrate subscription payment widgets on their website.”

    Zuora’s Subscription Economy Index report reveals that “the subscription economy grew by more than 435% in almost a decade.” UBS Wealth Management and Bernstein estimate that the subscription economy “will be worth $1.5 trillion by 2025.” The crypto industry is lagging behind “as most of its products and derivatives are sold outright, and merchants can’t easily tap into an existing ecosystem of infrastructure solutions.”

    Calypso Pay allows merchants “to sell goods and services by collecting recurring payments and subscriptions in cryptocurrencies, including USDT (including Polygon), USDC, BUSD (on Ethereum), FRAX, and DAI, in regular intervals.”

    Calypso Pay “takes care of subscription management and the underlying smart contracts and eliminates all the challenges.”

    Calypso Pay’s core platform “enables merchants to receive payments from clients quickly and securely and withdraw them to their wallets or bank accounts or directly pay their partners or workers.”

    Clients can “create invoices and/or use a payment widget to accept payments, initiate payouts for individuals, or make payments to up to 1000 recipients simultaneously.”

    Calypso Pay provides “one of the fastest transaction speeds with low commissions across the market – over the year, the platform processed $850 million, and its clients performed almost 1 million transactions.”

    The platform now “supports 14 main cryptocurrencies representing 80% of the global crypto market cap, including USDT (ERC-20, TRC-20, and Polygon), BTC, ETH, USDC, MATIC, BNB, and BUSD.”

    Alexey Korneev, Operations Director, said:

    “Subscriptions payments are the most reliable monetization option for businesses and creators. We had to solve multiple challenges in order to implement them in crypto and provide a smooth experience, and now we’re excited to offer this functionality to our customers”.

    This summer Calypso Pay has dramatically “increased the number of cryptocurrencies merchants can accept and has launched Tron Mass Payouts in partnership with WatchData, a platform that helps web3 developers seamlessly interact with the blockchain.”

    The solution “allows customers to quickly set up Tron payouts to tens of thousands of addresses while saving time and money on fees.”

    As noted in the update, Calypso Pay is “an all-in-one crypto processing & acquiring platform that lets you use crypto to receive payments from your clients and pay your partners and workers almost instantly and with minimal commissions.”

    The platform “enables payments with Bitcoin, Ethereum, Tron, MATIC, BSC, tokens such as USDT, various decentralized and algorithmic stablecoins, and many others.”

    It also allows “to exchange crypto and use SEPA and SWIFT bank transfers directly on the platform.”

    With Calypso Pay, you can “create invoices, make payments, initiate payouts, and analyze the financial health of your business in real time.”

    Crypto Processing Platform Calypso Pay Now Supports Subscriptions, Recurring Payments Republished from Source via

    Penn State Researchers Propose ‘ESFPNet,’ An Effective Deep Learning Network for Real-Time Lesion Segmentation in Autofluorescence Bronchoscopic Video


    The leading cancer mortality globally is Lung Cancer. A key objective for increasing lung cancer survival is discovering the illness early, allowing for the most effective treatment choices. Lung cancer develops from lesions in the bronchial epithelium of the lung mucosa. These bronchial lesions can progress to squamous cell lung cancer and assist in forecasting other lung cancers’ development. As a result, approaches for early diagnosis of bronchial lesions are critical for improving lung cancer patient treatment. Using bronchoscopy to image the airway epithelium during a regular airway exam is a noninvasive technique for clinicians to look for such lesions.

    Autofluorescence bronchoscopy is one of the most sensitive advanced bronchoscopic video procedures available today. It can efficiently distinguish growing bronchial lesions from the normal epithelium. Unfortunately, the current standard requires human inspection of an incoming AFB video stream, which is time-consuming and error-prone. While some research has looked toward computer-based lesion analysis approaches for AFB video frames, all of these studies have one or more limitations as follows: 

    1. Complicated image preprocessing is required before making lesion judgments.
    2. Do not offer reliable, real-time segmentation of aberrant lesion zones as a tool for finding prospective lesions.
    3. The approaches cannot interpret an input AFB video stream in real-time, rendering them inappropriate for making lesion determinations during a live bronchoscopic airway exam.

    Researchers believe this is the first time someone has used AFB video for automated real-time segmentation of bronchial lesions. Furthermore, their proposed efficient stage-wise feature pyramid (ESFP) encoder on Mixtransformer (MiT) with SOTA performances on public datasets demonstrates a significant capacity for medical picture segmentation.

    Their architecture is depicted in the figure below. It employs the Mix Transformer (MiT) encoder as the backbone and an efficient stage-wise feature pyramid (ESFP) decoder to create segmentation outputs. 


    GitHub has the official implementation of this paper.

    This Article is written as a research summary article by Marktechpost Staff based on the research paper 'ESFPNet: efficient deep learning architecture for real-time lesion segmentation in autofluorescence bronchoscopic video'. All Credit For This Research Goes To Researchers on This Project. Check out the paper and github link.
    Please Don't Forget To Join Our ML Subreddit

    Content Writing Consultant Intern at Marktechpost.

    Digital Investment Platform Stackwell Introduces Robo-Investing App


    Stackwell, the digital investment platform created to eliminate the racial wealth gap,  announced the release of its first product, a robo-investing app, which is currently “available for download in the App Store.”

    Designed for the Black community, the app focuses on “promoting accessibility, education, and support to help encourage individuals to participate in investing and build sustainable wealth.”

    Stackwell founder and CEO Trevor Rozier-Byrd said:

    “We are now uniquely positioned to help more people in the Black community leverage the power of the stock market to grow long-term sustainable wealth, and ultimately capture a massive market opportunity by becoming the trusted provider of end-to-end financial products and services for millions of Black consumers.”

    As noted in the update, the Stackwell app “provides users with all of the necessary tools to begin their investment journey.” The $1 monthly subscription fee and $10 investment minimum is “designed to increase accessibility across the income spectrum within the Black community.”

    The app is comprised of:

    • automated investment portfolios, recommended based on user-specific goals to help individuals invest with confidence;
    • in-app educational content that demystifies investing so users can grow their knowledge and their wealth; and
    • intentional design and science-based nudges to help users stay committed, stay consistent, and achieve their long-term wealth building goals and objectives.

    By narrowing the underinvestment gap, and increasing participation rates in the financial markets by the Black community, Stackwell will “help significantly lessen the brunt of the widening racial wealth gap that exists in our country today.”

    Omosefe Aiyevbomwan, Stackwell’s Vice President of Product, commented:

    “We identified the social, emotional and cultural barriers to entry, and built the platform to address each one head-on. Our model portfolios facilitate the process of getting started, and address lack of guidance and proximity to the markets; our financial wellness and education content answer for the lack of information; and the low-cost entry and subscription fees help assuage heightened risk aversion and concerns people have about investing without a safety net.”

    In conjunction with the app release today, Stackwell announced “a series of multi-year, strategic partnerships with several NBA and WNBA teams: The Detroit Pistons, The New Orleans Pelicans, The Minnesota Timberwolves and Minnesota Lynx, and The Washington Wizards and Washington Mystics.”

    These partnerships will “facilitate community engagement through unique programming and events designed to increase investment education and access to investment in the financial markets as a means to grow the wealth of the local Black communities.”

    Over the next three years, Stackwell and the teams will “develop co-branded content reaching several million people on local and national levels, open up over 3,500 new Stackwell accounts, all pre-seeded with initial funding for local program participants, and contribute over $250,000 back into the local communities to support the advancement and wealth building opportunities of Black students, entrepreneurs, creators, and small business owners, among others.”

    Stackwell, which recently announced the closing of its $3.5 million seed round, “led by Mike Gordon of Fenway Sports Group, Jeremy Sclar, CEO and Chairman of WS Development and The Kraft Group, will continue to build off of the momentum of its app release and NBA and WNBA partnerships to grow the ecosystem of support around the company, enabling execution against strategic company objectives, scaling of operations and driving impact in the Black community.”

    Digital Investment Platform Stackwell Introduces Robo-Investing App Republished from Source via

    DOJ: IRS Gains Court Order to Pursue Taxpayers Who Failed to Pay Taxes on Crypto


    The US Internal Revenue Service (IRS), the tax authority of the US, has received a Court Order to be able to summons records pertaining to US taxpayers who may have failed to report and pay taxes on crypto transactions.

    According to the Department of Justice (DOJ), US District Judge Paul G. Gardephe entered an order on September 22, 2022, authorizing the IRS to issue a John Doe summons requiring M.Y. Safra Bank to produce information about US taxpayers who may have failed to report to the IRS, and pay taxes on, cryptocurrency transactions.

    Specifically, the IRS summons seeks information about customers of SFOX, a crypto prime broker, who used banking services that M.Y. Safra Bank offered to SFOX customers engaged in cryptocurrency transactions, according to the DOJ.  SFOX is said to have over 175,000 registered users who have completed over $12 billion in transactions since 2015.

    U.S. Attorney Damian Williams issued the following statement:

     “Taxpayers are required to truthfully report their tax liabilities on their returns, and liabilities that arise from cryptocurrency transactions are not exempt.  The government is committed to using all of the tools at its disposal, including John Doe summonses, to identify taxpayers who have understated their tax liabilities by not reporting cryptocurrency transactions, and to make sure that everyone pays their fair share.”

    Charles P. Rettig, IRS Commissioner added:

    “The government’s ability to obtain third-party information on those failing to report their gains from digital assets remains a critical tool in catching tax cheats.  The court’s granting of the John Doe summons reinforces our ongoing, significant efforts to ensure that everyone pays their fair share. Taxpayers earning income from digital asset transactions need to come into compliance with their filing and reporting responsibilities.”

    The IRS believes that many digital asset transactions are not being reported correctly. The IRS states that investigations have identified at least ten US taxpayers who used SFOX’s services for cryptocurrency transactions but failed to report those transactions to the IRS.

    The IRS utilizes John Doe summonses to obtain information about possible violations of the internal revenue laws by individuals whose identities are unknown.

    On August 15, the IRS was authorized the US. District Court for the Central District of California to serve a John Doe summons on SFOX itself.

    A recent report distributed by John Reed Stark, a former SEC enforcement official, outlines further the activities being pursued by the DOJ and the IRS in regards to taxpayers believed not to have misreported, or simply not reported, crypto gains. Stark states that “Operation Hidden Treasure” is an ongoing project by the IRS that aims to uncover crypto tax evasion. He states that platforms like Circle and Kraken have been the target of John Doe Summonses.

    DOJ: IRS Gains Court Order to Pursue Taxpayers Who Failed to Pay Taxes on Crypto Republished from Source via

    AccessFintech Secures $60M via Series C Round Led By WestCap


    AccessFintech, the Fintech company evolving the capital markets operating model through data and workflow collaboration, announced the completion of its $60 million Series C funding round, led by WestCap.

    As the lead investor, WestCap claims that it “brings deep experience building and scaling transformational software for capital markets with companies including Ipreo, iLEVEL, BrokerTec, Tradeweb and SIMON Markets.”

    Additional investment was “secured from BNY Mellon and Bank of America, with further participation from Series B investors Dawn Capital, J.P. Morgan, Goldman Sachs and Citi Group.”

    The company’s latest investment round “follows a $20 million Series B and brings the total capital raised to $97 million since 2018.” The additional funds will be “used to accelerate AccessFintech’s growth as it expands its collaborative data management network to additional markets.”

    AccessFintech’s Board of Directors will be “joined by Kevin Marcus, Partner at WestCap and former President of Ipreo, and Caroline Butler, Global Head of Custody at BNY Mellon.”

    The funding “comes at the heels of rapid growth for the firm.”

    Synergy, AccessFintech’s data collaboration network, “has grown to include over 100 participants and expanded its asset class coverage to include derivatives and syndicated loans.”

    It has also “added to its securities offering by extending its lifecycle management through its new claims network and the launch of a ground-breaking predictive fails service.”

    The Synergy Network leverages “an innovative, cloud-based approach to data custody and management to reduce transactions costs and compress fees through collaboration and accelerated workflows, and helps organisations meet regulatory requirements.”

    It is used by “a growing number of leading global banks and broker-dealers, custodians, and asset managers.”

    Roy Saadon, CEO, AccessFintech, said:

    “We are fortunate to fuel our next stage of growth with partners that combine a highly regarded growth VC and global strategic investors. We share the vision of data being the catalyst for innovation and growth and the critical role AccessFintech plays as an essential provider of trusted data governance infrastructure. We are poised for a period of significant expansion and look forward to working with all our investors as we launch in additional markets.”

    Kevin Marcus, Partner, WestCap, said:

    “Behind an exceptional leadership team, AccessFintech has established itself as the technology partner trusted to solve long standing workflow challenges on behalf of the world’s leading financial institutions, creating a powerful data collaboration network, driving massive efficiency gains and unlocking pools of capital for its customers. AccessFintech’s sharing and pairing solution is the industry’s answer to complex issues such as the Central Securities Depositories regulation and penalty management, pre-matching and the transition to T+1 settlement.”

    Caroline Butler, CEO of Custody at BNY Mellon, remarked:

    “Democratizing data across market participants reduces operational friction and associated costs, and improves settlement, all of which are critical to our clients and the smooth functioning of the capital markets. This latest strategic venture reinforces our commitment to collaborating with leading fintechs and investing in emerging technology to transform the asset servicing space, as we look to expand and curate a digital ecosystem for clients.”

    AccessFintech Secures $60M via Series C Round Led By WestCap Republished from Source via

    Financial Stability Oversight Council , Re-Iterates that Digital Asset Report to be Published in October


    The Financial Stability Oversight Council held a meeting today, and part of the discussion revolved around digital assets.

    In the readout, the Council noted that Treasury staff provided an update on digital assets and the report that is being prepared following the recent Executive Order on Ensuring Responsible Development of Digital Assets.  The Council expects to issue the report in October 2022.

    Little additional information was provided.

    The Council includes representatives from all of the major financial regulators as well as several state regulators. The chair is Secretary of the Treasury, Janet Yellen.

    The recent Executive Order has received a certain amount of criticism from both sides of the fence with some policymakers slamming the document as enabling crypto fraud and others as “kicking the can down the road.” Treasury Staff leadership regarding digital assets has pursued a more balanced approach of recognizing the benefits of distributed ledger technology and the need for the US to lead innovation – albeit with a keen eye on investor protection and any potential for systemic risk.

    Financial Stability Oversight Council , Re-Iterates that Digital Asset Report to be Published in October Republished from Source via

    Kabbage Small Business Recovery Report Indicates Year Over Year Revenue Nearly Doubled. But Can it Continue?


    Kabbage, an online lender operated by American Express, has published its periodic Small Business Recovery Report – its 7th installment. The survey queried 550 small business leaders which indicate that year-over-year revenue, between July 2021 and July 2022, almost doubled (87%) while profits remained flat.

    Brett Sussman, Vice President Head of Sales & Marketing, Kabbage, claims that small businesses are not only surviving but flourishing even during historically high inflation and hiring challenges.

    Survey data indicates that overall profits among US small businesses have slightly declined by 4%  from July 2021 and July 2022. The data shows small businesses continue to anticipate future economic obstacles as 75% of respondents report feeling impacted by inflationary pressures and 56% expect pressures to last at least a year until the summer of 2023.

    The report shares:

    • 37% stated they plan to raise prices,
    • 22% aim to negotiate better deals with suppliers
    • 22% are cutting lower margin products and services from their offerings
    • 33% of small businesses plan to prioritize customer relations and strengthen customer loyalty to help increase future revenue.

    Small businesses are also adapting to the WFH movement by offering hybrid options. 49% of small businesses have started offering flexible work options to stay competitive in the labor market with hybrid work (27%) being the most popular option.

    The top two areas where SMEs are prioritizing effort are digital transformation (41%) and digital marketing (47%)

    While all of this displays small business resilience the economy is shifting rapidly as the US Federal Reserve battles the current administration as fiscal policy is at odds with monetary policy causing more problems. Inflation, once described as transitory, can no longer be ignored and the Fed has declared a need to stomp on inflation regardless of the impact it will have on the economy. The next Small Business Recovery Report will be far gloomier (and Kabbage should remove the term recovery).

    Kabbage Small Business Recovery Report Indicates Year Over Year Revenue Nearly Doubled. But Can it Continue? Republished from Source via

    Apple set to break average iPhone price record twice in coming months


    Apple is expected to break its average iPhone price record twice in the coming months as customers choose to buy costlier “Pro” models that boost the tech giant’s margins.

    Demand for the new iPhone 14 unveiled earlier this month is already robust enough to project that the global “average selling price” — or ASP — will rise to a record $892 in the September quarter and $944 in the December quarter, according to Counterpoint Research, a data provider, which bases its projections on consumer demand, market intelligence, and talks with suppliers.

    The current record is $873, achieved in the fourth quarter of last year.

    The average selling price of iPhones is a key Apple metric for Wall Street, as smartphone sales still account for roughly 50 per cent of the group’s revenues. The upward trajectory of these prices — from just $690 in late 2015 — is all the more significant given that Apple opted not to increase prices of its devices earlier this month at its flagship annual product release, a decision some analysts called the event’s biggest surprise.

    Driving the trend is the popularity of iPhone 14 Pro and Pro Max models, which feature a faster chip, a 48-megapixel camera and a new information hub called “Dynamic Island”.

    When research group Evercore ISI surveyed some 4,000 consumers this month it found that 56 per cent of those likely to buy an iPhone planned to buy a Pro model, versus 41 per cent last year. It anticipated that ASPs in the next year would be around $940, about 10 per cent higher than in the iPhone 13 cycle.

    “We are increasing our iPhone revenue estimates for the next four quarters on the back of stronger than expected ASPs,” Evercore analyst Amit Daryanani told clients.

    The figures suggest Apple finds itself in a much better position today than in January 2019, when it was forced to issue its first revenue warning in 16 years on the back of iPhone sales weakness in China.

    It has since taken greater market share in China thanks to the launch of 5G-enabled iPhones in 2020 and the downfall of Huawei, which has been hobbled from selling 5G phones following US sanctions.

    Between late 2020 and late 2021, Apple’s market share in China rose from 15.9 per cent to 23 per cent, according to Counterpoint.

    “Other Android manufacturers — Vivo, Oppo and Honor — all have entered the premium segment, but premium brand status cannot be built overnight,” said Counterpoint analyst Archie Zhang.

    Globally, the premium market of phones costing above $400 has outperformed overall sales for nine straight quarters, with Apple taking 57 per cent market share last quarter, Counterpoint data shows.

    The ultra-premium segment — phones priced above $1,000 — nearly doubled last quarter from a year ago, with Apple capturing 78 per cent of the market.

    The upward trajectory in the iPhone’s ASP reflects the success of a strategy Apple began in 2018 when it stopped reporting how many iPhones it was selling each quarter. That move had sparked “peak iPhone fears,” whereas Apple argued it was shifting focus from volume to revenues and margins.

    Today, analysts believe Apple is in another transition to shift from ASPs towards “lifetime user value” — a strategy of increasing revenues from its more than 1bn iPhone users with an expanding array of services.

    In late 2020 the company launched Apple One, offering a bundle of Apple services including Music and iCloud at one discounted price. The next logical step is for an iPhone itself to be part of the bundle, enabling users to pay a monthly fee in perpetuity in exchange for services and a new smartphone every year or two, without incurring any upfront cost.

    Apple set to break average iPhone price record twice in coming months Republished from Source via

    Fintech Raisin Moves into Crypto with New Service “Raisin Crypto,” Kicks Off in Germany


    Raisin, a European savings optimizer service that also operates in the US, has announced Raisin Crypto with the new offering first being launched in Germany under the Weltsparen brand. Raisin operates in the US under the SaveBetter brand.

    The Berlin-based Fintech crypto offering is designed to make investments in a passive portfolio that covers the most important cryptocurrencies.  Raisin Crypto aims to appeal to people without prior knowledge of cryptocurrencies. Users can invest in crypto ETNs [exchange traded notes] for a one-off investment of €500 or a savings plan of €50 a month.

    In a corporate statement, Raisin said that users can invest in a portfolio of the most important crypto stocks, as well as certain cryptocurrencies via Raisin’s German platform WeltSparen. The portfolio represents the largest possible part of the crypto market and seeks to be less susceptible to market fluctuations than single tokens. By incorporating crypto, Raisin describes WeltSparen’s offering as a “holistic platform for savings, investments, and retirement provision.”

    The company states that Raisin Crypto adheres to its investment philosophy of providing an easy to understand, low cost and diversified products. This is different from speculative trading. Raisin Crypto uses a passive investment strategy that allows customers to invest in the crypto market even without experience. At the moment, individual digital assets available include, Bitcoin, Ethereum, Cardano, Solana, Polkadot, Polygon, and Avalanche.

    Raisin states that a broadly diversified Crypto investment can enable superior profits and reduce risk at the same time.  An automated rebalancing takes place every quarter to restore the weighting in the portfolio. The costs for Raisin Crypto amount to 1.5% of the portfolio value per year. Performance and management fees as well as custody and transaction costs do not apply.

    Raisin Crypto invests via an exchange-traded note (ETN). Investors can profit from rising shares without having to hold coins themselves or a wallet for managing cryptocurrencies. After concluding the product, customers can easily and transparently track the performance of the individual cryptocurrencies in their WeltSparen account. Investing in Raisin Crypto is possible f

    Kim Felix Fomm, Chief Investment Officer at Raisin, commented:

    “Cryptocurrencies are one of the most volatile asset classes, which leads to an asymmetric risk-reward ratio: If you invest a small amount, the potential loss is limited to this amount, but the participation in the return is unlimited. Therefore, we consider an allocation of up to five percent in cryptocurrencies in an appropriately diversified portfolio to be sensible. Those who decide to do so must be able to withstand strong price fluctuations. Raisin Crypto is our alternative to short-term trading in single cryptocurrencies, which is usually unprofitable for private investors. We offer a broadly diversified, user-friendly solution that works as a solid investment and can yield a good return in the long run.”

    Raisin did not reveal its intent for other European markets or North America.

    In Germany, in addition to savings products the company offers ETF-based investment and retirement products as well as Private Equity investments. Raisin works with over 400 banks and financial service providers from more than 30 countries and reports over 750,000 direct customers.

    Fintech Raisin Moves into Crypto with New Service “Raisin Crypto,” Kicks Off in Germany Republished from Source via

    Google’s New Open-Source Project, ‘SayCan,’ is an AI Tool That Uses a Large Language Model to Plan Sequences of Robotic Actions to Achieve a User-Specified Goal


    Various semantic information about the world can be encoded by large language models (LLM). Nevertheless, they can often generate responses that, while logically sound, would not be helpful for controlling a robot. The lack of contextual grounding in language models is a severe disadvantage. A language model might provide a fair narrative in response to a user’s request for instructions on cleaning up a spill, but a robot performing this task in a specific context might not find it helpful. As a result, it is challenging to use them in real-world contexts for decision-making.

    In a recent work titled “Do As I Can, Not As I Say: Grounding Language in Robotic Affordances,” researchers from Google’s Robotics team offered an innovative strategy. This paper presents SayCan, a robot control approach that plans a series of robotic operations to accomplish a user-specified goal using an LLM. The method employs prompt engineering to translate the user’s input—in this case, a request for assistance in cleaning up milk—into a dialogue asking the robot to deliver the user a sponge. According to experimental analyses, SayCan generated the right action sequence 84% of the time.

    The approach’s premise is that the language model may provide high-level semantic information about the activity, while the robot can serve as its “hands and eyes.” The work of the researchers shows how low-level tasks can be combined with LLMs in a way that the language model provides high-level knowledge about the methods for carrying out complex and temporally extended instructions. In contrast, value functions associated with these tasks provide the grounding required to connect this knowledge to a specific physical environment. The method was tested on various robotic jobs, demonstrating its viability for executing long-horizon, abstract, natural language commands on a mobile manipulator.

    The raw user input was preceded by a chain of thought prompt that included 17 example inputs and their corresponding plans to enhance the LLM’s capacity to plan a series of activities in SayCan. The text description of a skill (the probability that skill is useful for the instruction) and its value function output (the probability of successfully executing said skill) can be used to choose the best action in the plan sequence. This is possible because LLM outputs a probability distribution over text tokens for the following item in a sequence. 

    A robot from Everyday Robots, who collaborated with Google on this project, was given a list of 101 commands to follow, ranging from “bring me a fruit” to “I spilled my coke on the table, throw it away and bring me something to clean,” in order to test SayCan. PaLM and FLAN are just two of the LLMs that Google integrated SayCan with. With a planning success rate of 84% and an execution success rate of 74%, PaLM-SayCan outperformed FLAN-SayCan, which had success rates of 70% and 61%, respectively. The team noticed that PaLM-SayCan had trouble with instructions containing a negative, but they also noted that this is a typical problem with LLMs in general. 

    The impressive development made by PaLM-SayCan opens up new study horizons. This study explains how a model can be used to solve reasoning problems by utilizing chain of thought reasoning and how new skills can be included in the system. Additionally, it demonstrates that the system can handle multilingual inquiries even if it was not intended. The researchers also think that PaLM-interpretability SayCan’s enables secure user interactions with robots in the real world. 

    The researchers want to understand further how data from the robot’s real-world experience could be used to enhance the language model and to what extent natural language is the appropriate ontology for programming robots as they explore future paths for this work. In order to give academics a helpful tool for upcoming research that blends robotic learning with sophisticated language models, Google Research has also open-sourced a robot simulation setup. An open-source desktop version of SayCan is now available on GitHub. 

    This Article is written as a research summary article by Marktechpost Staff based on the research paper 'Do As I Can, Not As I Say:
    Grounding Language in Robotic Affordances'. All Credit For This Research Goes To Researchers on This Project. Check out the paper, project, github link and reference article.
    Please Don't Forget To Join Our ML Subreddit

    Khushboo Gupta is a consulting intern at MarktechPost. She is currently pursuing her B.Tech from the Indian Institute of Technology(IIT), Goa. She is passionate about the fields of Machine Learning, Natural Language Processing and Web Development. She enjoys learning more about the technical field by participating in several challenges.

    Bambu Offers Wealth Management Predictive Planning Solution for Microsoft Cloud for Financial Services


    Bambu, a global wealth technology provider, announced the launch of their Wealth Management Predictive Planning solution for Microsoft Cloud for Financial Services and Microsoft Dynamics 365 Sales.

    The service will “enable Relationship Managers (RMs), Financial Advisors, Agents and firms in the financial sector to identify and prioritize their clients’ needs through predictive analytics, empowering them to make informed financial decisions for the future.”

    To build and solidify trusted relationships between users and their clients while increasing sales productivity, Wealth Management for Microsoft Cloud for Financial Services – a one-of-a-kind Wealth Management predictive planning solution – along with its counterpart, Wealth Management for Dynamics 365 Sales, “were designed with an innovative proactive approach in mind.”

    The system will “enable users to predict and analyze key life events along with the ability to suggest achievable financial goals and solutions for future financial planning.”

    Powered by A.I. and through the adoption of the Hidden Markov Model (HMM) and Generative Adversarial Network (GAN) Machine Learning models, “both solutions will be able to generate thousands of timelines of future events and outlines based on demographic profiles and the four major aspects of life – marriage, family, career, and financial situation.”

    Aki Ranin, Co-founder and COO of Bambu, said:

    “It is vital for us to move the game from reactionary, aspirational planning to proactive, realistic, and predictive planning. We hope to revolutionize and be at the forefront of the playing field to make it much easier for financial planning to be a priority.”

    Ajay Kamath, Commercial Director, at Bambu, stated:

    “Bambu wants to be the leader in the B2B wealth tech space. This relationship means a lot to us, not only from a brand perspective but also from a product scale standpoint. The opportunity to co-sell with Microsoft will help us progress toward our goal to be a leader in the B2B wealth technology space.’

    Toby Bowers, General Manager, Industry, Applications & Data Marketing at Microsoft, remarked:

    “We welcome Bambu to Microsoft’s Business Application ecosystem and its Wealth Management offering integrated with Microsoft Cloud for Financial Services and Microsoft Dynamics 365 Sales. By making it easier for relationship managers, financial advisors and agents to access their data in one place, we are significantly reducing roadblocks that lead to friction and reduced ROI.”

    Bambu’s Wealth Management for Microsoft Cloud for Financial Services and Wealth Management for Dynamics 365 Sales is now “available on Microsoft AppSource.”

    Bambu Offers Wealth Management Predictive Planning Solution for Microsoft Cloud for Financial Services Republished from Source via

    NVIDIA Introduces the NVIDIA IGX Platform for Medical Edge AI Use Cases


    Organizations across various sectors are exploring ways to increase their use of automation to boost productivity, efficiency, and security. Computer programs can make this process fast by learning to recognize patterns and reliably carry out the same tasks repeatedly. However, the reality is not deterministic, and the breadth of human activity encompasses various tasks and contexts that rules and programs cannot adequately capture.

    Edge AI refers to the practice of doing AI computation at the network’s periphery, closer to the user and closer to the actual data, rather than at a centralized location like a cloud service provider’s server farm or a private company’s data center.

    Edge AI has advanced to the point where machines and devices may now function with the “intelligence” of human cognition wherever they may be. Intelligent apps powered by AI can adapt to new situations and learn to execute the same or similar tasks successfully.

    With the release of the NVIDIA IGX platform, NVIDIA has brought cutting-edge safety and security to intelligent machines and human-machine collaboration, making it ideal for application in medical edge AI use cases.

    Secure, low-latency AI inference is provided by IGX, a hardware and software platform, to satisfy the clinical requirement for rapid insights from various devices and sensors for medical applications, including robotic-assisted surgery and patient monitoring.

    NVIDIA Clara Holoscan is a domain-specific platform that facilitates connectivity between edge, on-premises data center, and cloud services, and the IGX platform supports it. This synergy expedites the creation of cutting-edge artificial intelligence (AI) technologies that can be used in the operating room.

    Activ Surgical, Moon Surgical, and Proximie, three of the world’s leading medical-device firms, have all chosen to utilize NVIDIA Clara Holoscan on the IGX platform to power their surgical robotics systems. They are all participants in NVIDIA Inception, a worldwide initiative designed to accelerate the development of innovative new businesses. Currently, Clara Holoscan is being used by over 70 medical device manufacturers, hospitals, and startups to help speed up the process of deploying AI solutions in clinical settings.

    NVIDIA IGX Orin, the world’s most powerful, compact, and energy-efficient artificial intelligence (AI) supercomputer for medical devices, drives the NVIDIA IGX platform. To facilitate the transition from clinical trials to real-world deployment, IGX incorporates medically certified industrial-grade components. 

    Some of the earliest companies to develop NVIDIA IGX-based devices for the medical device market include ADLINK, Advantech, Dedicated Computing, Kontron, Leadtek, MBX, Onyx, Portwell, Prodrive Technologies, and YUAN.


    Tanushree Shenwai is a consulting intern at MarktechPost. She is currently pursuing her B.Tech from the Indian Institute of Technology(IIT), Bhubaneswar. She is a Data Science enthusiast and has a keen interest in the scope of application of artificial intelligence in various fields. She is passionate about exploring the new advancements in technologies and their real-life application.

    Instagram allowed self-harm images so people could ‘cry for help’, inquest hears


    A senior Meta executive told a British inquest the company had allowed “graphic” images of self-harm on its Instagram site at the time a teenager died by suicide because it wanted to enable users to “cry for help”.

    Molly Russell from Harrow, London, died in November 2017 after viewing a large volume of posts on sites such as Meta-owned Instagram, and Pinterest, related to anxiety, depression, suicide and self-harm.

    Meta’s head of health and wellbeing policy, Elizabeth Lagone, told the North London coroner’s court on Friday that graphic images Instagram allowed users to share at the time of Russell’s death, could have been “cries for help” and the platform wanted people to “seek community”.

    “Graphic promotion or encouragement [of suicide or self-harm] was never allowed,” she said, but added that “silencing [a poster’s] struggles” could cause “unbelievable harm”.

    She said the issues were “complicated” and that expert understanding had evolved in recent years.

    The court was shown a series of video clips that Russell had liked or saved on Instagram before she died, including close-ups of individuals cutting their wrists with razor blades, that senior coroner Andrew Walker said were “almost impossible to watch”. 

    The clips included close-up shots of people self-harming, falling from buildings and swallowing handfuls of pills, often spliced with loud music and negative messages. It was unclear whether they displayed real events or were taken from film and TV.

    Walker said the content “appears to glamorise harm to young people” and was “of the most distressing nature”.

    Lagone said Instagram had changed its policy in 2019 after experts advised it that graphic self-harm imagery could encourage users to hurt themselves. The company previously removed posts that glorified, encouraged or promoted self-harm but not posts that could have enabled users to admit their struggles and support each other.

    After Russell’s death, experts advised the company that “some graphic images . . . could have the potential to promote self-injury,” according to part of Lagone’s witness statement read out in court.

    When asked if Meta had undertaken research into the impact of self-harm content on users, Lagone said she was not aware of any and that it would have been difficult to conduct. “The impact of certain material can affect people in different ways at different times . . . It’s really complicated,” she said.

    Molly Russell’s father, Ian Russell, told the inquest this week that social media algorithms had pushed his daughter towards disturbing posts and contributed to her death. He told the court that “social media helped kill my daughter”.

    Instagram had recommended accounts to Molly Russell that included some related to depression and suicidal feelings.

    Molly Russell had also been recommended content about depression by Pinterest, the inquest heard this week, including “ten depression pins you might like”. She continued to receive emails from Pinterest after her death, including one entitled “new ideas for you in depression”.

    On Thursday a senior Pinterest executive admitted to the inquest that the site had not been safe at the time of Molly Russell’s death and was still “imperfect”, in spite of updates to its rules.

    When asked by the Russell family’s barrister, Oliver Sanders KC, if Instagram’s policies had been “inadequate” when Molly Russell died, Lagone said: “We concluded that we needed to expand the policies and we did so.”

    The hearing comes as the passage through parliament of the online safety bill, which aims to compel internet companies to keep their platforms safe, has been paused. Liz Truss, the new prime minister, is said to be considering relaxing a clause that is controversial among tech lobbyists which would make platforms responsible for removing content that was “legal but harmful”, such as bullying.

    Instagram allowed self-harm images so people could ‘cry for help’, inquest hears Republished from Source via

    Regulators Try to Tighten Their Grip on Crypto


    Crypto Market Musings

    On Wednesday, the Fed raised interest rates for the third time this year by 75 basis points. Both the stock and crypto markets responded quickly to the aggressive rate hike. As of this writing, bitcoin is down 2.6% and trading below $19,000. Ethereum fell by 6.6% and is trading below $1,300. Last week ethereum fell by 16% and bitcoin fell by 6.3% — in part because the Consumer Price Index (CPI) came in higher than expected at 8.3%.


    Bitcoin (BTC)

    $ 19,100.94


    Ethereum (ETH)

    $ 1,344.22

    This probably won’t be the last time the Fed raises interest rates in an attempt to bring inflation back to 2%. Further rate hikes could trigger a recession, which would drive the crypto markets down even more. I think it is going to take a while for it to recover. Smaller crypto coins might even shut down entirely.

    I am bearish on the crypto market for this year. And that’s not only because of the “bad” economy. It’s also because the SEC is tightening regulations and the Fed is taking more notice of the crypto market.

    Get Early Investing into your inbox

    Become a smarter investor in startups, crypto and cannabis by subscribing to our FREE newsletter filled with market research, trends and expert analysis.

    What Yasmin Is Thinking About

    One thing I think will play a big role in determining future crypto prices is crypto market regulations. As crypto has become more popular and mainstream, policymakers have been getting more serious about regulating the market.

    As I was writing this article, I learned that Colorado has become the first state to accept crypto as payment for taxes through PayPal. This is big news. When crypto started, many people thought of it (and many still think of it) as a Ponzi scheme… and now a state government is integrating it into its tax payment system. This is a major step for crypto as it is gaining more recognition and trust. However, this also creates problems as it means the government and regulators are now more aware of the space and more likely to create regulations that could stifle the crypto market’s growth. In a paper published back in August, the Fed acknowledged the importance of decentralized finance (DeFi) but also argued that DeFi and the crypto market should be regulated more.

    And to no one’s surprise, this is what the SEC has long been after. On Monday, the SEC sued crypto influencer Ian Balina. And in a 23-page court filing, it stated that Ethereum nodes are more clustered in the U.S. than anywhere else — and the clustering of those nodes means Ethereum transactions happen inside the U.S. More than 40% of Ethereum nodes do operate within the U.S. But whether that means all Ethereum transactions happen in the U.S. and whether that gives the U.S. jurisdiction over all Ethereum transactions is a novel and untested legal argument. But it does show that the SEC is grasping for ways to establish regulatory oversight over crypto.

    Last week, SEC head Gary Gensler stated that cryptos that allow users to stake coins pass the Howey Test, a test created by the Supreme Court to decide whether a transaction is a security. For coins that use staking, Gensler said, “the investing public is anticipating profits based on the efforts of others.” Gensler believes that coins that use proof of stake — which now include Ethereum — could therefore be considered securities. Although Gensler did not specify any crypto while discussing proof of stake, the statement came the same day that Ethereum transitioned from proof of work to proof of stake. The Merge has clearly been on Gensler’s mind too. 

    There seems to be a split among crypto enthusiasts on whether the SEC should regulate the crypto space. I personally lean against SEC regulation of the crypto market. Although SEC oversight could make it more secure, I also think tight regulations and policies could ruin many projects and could possibly put too many restrictions on investors, which in turn would steer them away from investing in the space. This has already happened outside the U.S. The Ontario Securities Commision just introduced restrictions on non-accredited investors that limits their altcoin purchases to $30,000 Canadian dollars a year. I believe these small changes will definitely have a bigger impact on crypto prices down the road.

    And Finally…

    Recently — thanks to a friend — I have gotten a taste of how powerful the metaverse could become. At first I was skeptical. When I used a virtual reality (VR) headset before, all I experienced was headaches and nausea. But with my friend’s super gaming computer connected to the Oculus Quest 2 VR headset, I was able to experience what’s considered the best VR game made so far — “Half-Life: Alyx.” 

    I was blown away. The game changed my view of what the metaverse could become. The graphics quality was on a completely different level. The game is very detailed and even takes a player’s height into consideration. (I had to stand on my tiptoes sometimes to reach items up in cabinets.) The game lets you do a lot of things you can do in real life. It felt like a parallel universe. 

    I was privileged to experience it, and I encourage everyone to do so if they get the chance. This time, instead of dealing with nausea or a headache, I was in awe and fear of how real everything felt, especially as the enemies in the game were getting close to me. After I took off the VR headset, all I could think was “now I get why people are excited about the metaverse.”

    I understand that “Half-Life” is not exactly the metaverse, as the metaverse is more about how people interact with each other, and it means different things to different people. However, the game is a very good example of the huge potential the space has with the right graphics and immersive and interactive VR technology. As VR technology advances in different areas, I have no doubt more people will use it and understand it just like I did. 

    Meta is having its VR Connect conference on October 11 to reveal its new Oculus Quest Pro VR headset. If a 2-year-old Quest made me change my perspective about the metaverse, then I can only imagine what a Quest Pro would do. Take my money, Meta!

    I think with the release of Meta’s new headset, people will steadily become more aware of the metaverse’s potential. And I think this awareness will affect the crypto market too. In the future, the metaverse might not exist without crypto as it becomes its primary digital payment. Soon we could see many metaverse crypto projects take off as they attract more supporters who experience the “aha” moment like I did. If you are interested in what crypto metaverse project we have eyes on, check out Vin Narayanan’s latest metaverse crypto recommendation for Crypto Asset Strategies members. (If you’re not already a member, you can sign up here.)

    Regulators Try to Tighten Their Grip on Crypto Republished from Source via

    Leaked Audio Emerges of Celsius’ Bankruptcy Plan


    Celsius, a crypto company that collapsed into bankruptcy in spectacular fashion, has had a discussion pertaining to its failure leaked on the internet.

    Posted by Tiffany Fong, she has also included a transcription of the audio as well as her opinion of the discussion. Apparently, there is a plan to issue an IOU token to make investors whole. Take it for what it is worth. You can listen to the audio below which was apparently shared with Fong on  September 1, 2022, anonymously.


    Leaked Audio Emerges of Celsius’ Bankruptcy Plan Republished from Source via

    Whitepaper: Modern AppSec in financial services



    Whitepaper: Modern AppSec in financial services – FinTech Futures Modern AppSec in Financial Services