Page 52«..1020..51525354..6070..»

Crypto Whales Are Accumulating This Altcoin Amid Crypto Market Crash – Coinpedia Fintech News

The cryptocurrency industry has experienced a downturn in recent weeks, with midterm optimism fading. Chainlink (LINK) has been particularly affected, seeing its price drop over 21% in the past month, settling around $14.33 on Wednesday during the early London session.

Despite recent declines, the crypto market is expected to rebound as more investors enter the space. The Federal Reserve is likely to maintain a dovish stance in the coming quarters, with inflation easing and a general election on the horizon, signaling potential interest rate cuts. Meanwhile, European and Canadian governments have already begun reducing rates, setting a positive tone for the industry.

On-chain data analysis from Lookonchain reveals significant activity in the Chainlink ecosystem. Over the past week, 54 new wallets withdrew a total of 2.08 million LINK, worth approximately $30.28 million, from Binance. The Chainlink network, supported by both institutional investors and retail traders, boasts over 720,000 holders with non-zero balances, highlighting its robust and growing community.

Chainlink is a vital player in providing oracle data for Web3 and smart contract developers. Institutional investors aiming to build Web3 projects, particularly those involving real-world asset (RWA) tokenization, have increasingly turned to Chainlinks services. The Cross-Chain Interoperability Protocol (CCIP) by Chainlink is already in use by numerous institutions to develop scalable and interoperable applications.

In recent news, Chainlink has announced partnerships with Fidelity International and Sygnum. These collaborations aim to provide NAV data on-chain for Fidelity Internationals $6.9 billion money market fund, underscoring Chainlinks growing influence in the financial sector.

Chainlinks price has been in a megaphone pattern of weekly consolidation since the start of the year, even as Bitcoin has surged to new heights. Popular crypto analyst Michal van de Poppe suggests that LINK is on the brink of a significant bullish breakout in the near term.

Additionally, a reversal in Bitcoin dominance could herald the much-anticipated altseason, potentially boosting LINK and other altcoins.

Also Read: Top Altcoins Below $1 To Invest Before the Revival of a Fresh Bullish Spell

Dont miss the boat! With strong fundamentals and a bullish outlook, Chainlink could be your next big crypto win.

Here is the original post:
Crypto Whales Are Accumulating This Altcoin Amid Crypto Market Crash - Coinpedia Fintech News

Read More..

The quantum physics behind fireworks displays – Big Think

This Thursday, July 4, 2024, is remarkable for a number of reasons. It happens to bejust one day before aphelion: the day where the Earth is at its most distant from the Sun as it revolves through the Solar System in its elliptical orbit. Its the 248th anniversary of when the United States officially declared independence from, and war on, the nation of Great Britain. And it marks the annual date when the wealthiest nation in the world sets off more explosivesin the form of fireworksthan any other.

Whether youre an amateur hobbyist, a professional installer, or simply a spectator, fireworks showsare driven by the same laws of physicsthat govern all of nature. Individual fireworks all contain the same four component stages: launch, fuse, burst charges, and stars. Without quantum physics, not a single one of them would be possible. Heres the science behind how every component of these spectacular shows works.

The anatomy of a firework consists of a large variety of elements and stages. However, the same four basic elements are the same across all types and styles of fireworks: the lift charge, the main fuse, a burst charge, and stars. Variations in the diameter of the launch tube, the length of the time-delay fuse, and the height of the fireworks are all necessary to ignite the stars with the proper conditions during the break.

The start of any firework is the launch aspect: the initial explosion that causes the lift. Ever sincefireworks were first inventedmore than a millennium ago, the same three simple ingredients have been at the heart of them: sulfur, charcoal, and a source of potassium nitrate. Sulfur is a yellow solid that occurs naturally in volcanically active locations, while potassium nitrate is abundant in natural sources like bird droppings or bat guano.

Charcoal, on the other hand, isnt the briquettes we commonly use for grilling, but the carbon residue left over from charring (or pyrolyzing) organic matter, such as wood. Once all the water has been removed from the charcoal, all three ingredients can be mixed together with a mortar and pestle. The fine, black powder that emerges is gunpowder, already oxygen-rich from the potassium nitrate.

The three main ingredients in black powder (gunpowder) are charcoal (activated carbon, at left), sulfur (bottom right) and potassium nitrate (top right). The nitrate portion of the potassium nitrate contains its own oxygen, which means that fireworks can be successfully launched and ignited even in the absence of external oxygen; they would work just as well on the Moon as they do on Earth.

With all those ingredients mixed together, theres a lot of stored energy in the molecular bonds holding the different components together. But theres a more stable configuration that these atoms and molecules could be rearranged into. The raw ingredientspotassium nitrate, carbon, and sulfurwill combust (in the presence of high-enough temperatures) to form solids such as potassium carbonate, potassium sulfate, and potassium sulfide, along gases such as carbon dioxide, nitrogen, and carbon monoxide.

All it takes to reach these high temperatures is a small heat source, like a match. The reaction is a quick-burning deflagration, rather than an explosion, which is incredibly useful in a propulsion device. The rearrangement of these atoms (and the fact that the fuel contains its own oxygen) allows the nuclei and electrons to rearrange their configuration, releasing energy and sustaining the reaction. Without the quantum physics of these rearranged bonds, there would be no way to release this stored energy.

The Macys Fourth of July fireworks celebration that takes place annually in New York City displays some of the largest and highest fireworks you can find in the United States of America and the world. This iconic celebration, along with all the associated lights and colors, is only possible because of the inescapable rules of quantum mechanics.

When that first energy release occurs, conventionally known as the lift charge, it has two important effects.

The upward acceleration needs to give your firework the right upward velocity to get it to a safe height for explosion, and the fuse needs to be timed appropriately to detonate at the peak launch height. A small fireworks show might have shells as small as 2 inches (5 cm) in diameter, which require a height of 200 feet (60 m), while the largest shows (like the one by the Statue of Liberty in New York) have shells as large as 3 feet (90 cm) in diameter, requiring altitudes exceeding 1000 feet (300 m).

Different diameter shells can produce different sized bursts, which require being launched to progressively higher altitudes for safety and visibility reasons. In general, larger fireworks must be launched to higher altitudes, and therefore require larger lift charges and longer fuse times to get there. The largest fireworks shells exceed even the most grandiose of the illustrations in this diagram.

The fuse, on the other hand, is the second stage and will be lit by the ignition stage of the launch.Most fusesrely on a similar black powder reaction to the one used in a lift charge, except the burning black powder core is surrounded by wrapped textile coated with either wax or lacquer. The inner core functions via the same quantum rearrangement of atoms and electron bonds as any black powder reaction, but the remaining fuse components serve a different purpose: to delay ignition.

The textile material is typically made of multiple woven and coated strings. The coatings make the device water resistant, so they can work regardless of weather. The woven strings control the rate of burning, dependent on what theyre made out of, the number and diameter of each woven string, and the diameter of the powder core. Slow-burning fuses might take 30 seconds to burn a single foot, while fast-burning fuses can burn hundreds of feet in a single second.

The three main configurations of fireworks, with lift charges, fuses, burst charges, and stars all visible. In all cases, a lift charge launches the firework upward from within a tube, igniting the fuse, which then burns until it ignites the burst charge, which heats and distributes the stars over a large volume of space.

The third stage, then, is the burst charge stage, which controls the size and spatial distribution of the stars inside. In general the higher you launch your fireworks and the larger-diameter your shells are, the larger your burst charge will need to be to propel the insides of the shell outward. In general, the interior of the firework will have a fuse connected to the burst charge, which is surrounded by the color-producing stars.

Theburst chargecan be as simple as another collection of black powder, such as gunpowder. But it could be far more complex, such as the much louder and more impressiveflash powder, or a multi-stage explosive that sends stars in multiple directions. By utilizing different chemical compounds that offer different quantum rearrangements of their bonds, you can tune your energy release, the size of the burst, and the distribution and ignition times of the stars.

Differently shaped patterns and flight paths are highly dependent on the configuration and compositions of the stars inside the fireworks themselves. This final stage is what produces the light and color of fireworks, and is where the most important quantum physics comes into play.

But the most interesting part is that final stage: where the stars ignite. The burst is what takes the interior temperatures to sufficient levelsto create the light and colorthat we associate with these spectacular shows. The coarse explanation is that you can take different chemical compounds, place them inside the stars, and when they reach a sufficient temperature, they emit light of different colors.

This explanation, though, glosses over the most important component: the mechanism of how these colors are emitted. When you apply enough energy to an atom, or molecule, you can excite or even ionize the electrons that conventionally keep it electrically neutral. When those excited electrons then naturally cascade downward in the atom, molecule, or ion, they emit photons, producing emission lines of a characteristic frequency. If they fall in the visible portion of the spectrum, the human eye is even capable of seeing them.

The traditional model of an atom, now more than 100 years old, is of a positively charged nucleus orbited by negatively charged electrons. Although the outdated Bohr model is where this picture comes from, we can arrive at a more accurate description simply by considering the electrons quantum uncertainty.

What determines which emission lines an element or compound possesses? Its simply the quantum mechanics of the spacing between the different energy levels inherent to the substance itself. For example, heated sodium emits a characteristic yellow glow, as it has two very narrow emission lines at 588 and 589 nanometers. Youre likely familiar with these if you live in a city, as most of those yellow-colored street lamps you see are powered by elemental sodium.

As applied to fireworks, there are a great variety of elements and compounds that can be utilized to emit a wide variety of colors. Different compounds of Barium, Sodium, Copper, and Strontium can produce colors covering a huge range of the visible spectrum, and the different compounds inserted in the fireworks stars are responsible for everything we see. In fact,the full spectrum of colors can be achievedwith just a handful of conventional compounds.

The interior of this curve shows the relationship between color, wavelength, and temperature in chromaticity space. Along the edges, where the colors are most saturated, a variety of elements, ions, and compounds can be shown, with their various emission lines marked out. Note that many elements/compounds have multiple emission lines associated with them, and all of these are used in various fireworks. Because of how easy it is to create barium oxide in a combustion reaction, certain firework colors, such as forest green and ocean green, remain elusive.

Whats perhaps the most impressive about all of this is that the color we see with the human eye is not necessarily the same as the color emitted by the fireworks themselves. For example, if you were to analyze the light emitted by a violet laser, youd find that the photons emerging from it were of a specific wavelength that corresponded to the violet part of the spectrum.

The quantum transitions that power a laser always result in photons of exactly the same wavelength, and our eyes see them precisely as they are, with the multiple types of cones we possess responding to that signal in such a way that our brain responds to construct a signal thats commensurate with the light possessing a violet color.

A set of Q-line laser pointers showcase the diverse colors and compact size that now are commonplace for lasers. By pumping electrons into an excited state and stimulating them with a photon of the desired wavelength, you can cause the emission of another photon of exactly the same energy and wavelength. This action is how the light for a laser is first created: by the stimulated emission of radiation.

But if you look at that same color that appears as violet not from a monochromatic source like a laser, but from your phone or computer screen, youll find that there are no intrinsically violet photons striking your eyes at all! Instead,as Chad Orzel has noted in the past,

Travel the Universe with astrophysicist Ethan Siegel. Subscribers will get the newsletter every Saturday. All aboard!

Our eyes construct what we perceive as color from the response of three types of cells in our retina, each sensitive to light of a particular range of colors. One is most sensitive to blue-ish light (short wavelength), one is most sensitive to red light (long wavelength), and the third to a sort of yellow-green. Based on how strongly each of these cells responds to incoming light, our brains construct our perception ofcolor.

In other words, the key to producing the fireworks display you want isnt necessarily to create light of a specific color that corresponds to a specific wavelength, but rather to create light that excites the right molecules in our body to cause our brain to perceive a particular color.

A violet laser emits photons of a very particular, narrow wavelength, as every photon carries the same amount of energy. This curve, shown in blue, emits violet photons only. The green curve shows how a computer screen approximates the same exact violet color by using a mix of different wavelengths of light. Both appear to be the same color to human eyes, but only one truly produces photons of the same color that our eyes perceive.

Fireworks might appear to be relatively simple explosive devices. Pack a charge into the bottom of a tube to lift the fireworks to the desired height, ignite a fuse of the proper length to reach the burst charge at the peak of its trajectory, explode the burst charge to distribute the stars at a high temperature, and then watch and listen to the show as the sound, light, and color washes over you.

Yet if we look a little deeper, we can understand how quantum physics underlies every single one of these reactions. Add a little bit extrasuch as propulsion or fuel inside each starand your colored lights can spin, rise, or thrust in a random direction. Make sure you enjoy your fourth of July safely, but also armed with the knowledge that empowers you to understand how the most spectacular human-made light show of the year truly works!

A version of this article first appeared in 2022. Happy 4th of July, everyone!

Read the original:

The quantum physics behind fireworks displays - Big Think

Read More..

Scientists Tracked Earth’s Rotation With Frankly Unbelievable Precision – Popular Mechanics

In a paper recently published in the journal

Interferometry is a technique involving displaced and refracted lightthe patterns created reveal what interfered in the first place. It was originally conceived of to help prove or disprove the idea that all the air around us was filled with a wild proposed material called luminiferous ether, but now its used in a lot of different scientific disciplines. Most interferometers use mirrors, curved lenses, and more to bend and recapture light waves, but some may be acoustic, using a crystal to resonate through a cloud of liquid or gas.

These setups predate the study of quantum mechanics, but the researchers from the Vienna Center explain in their paper how quantum interferometry improves on previous forms. [T]he enhanced sensitivity of quantum interferometers, they wrote, opens up opportunities for precision measurements that can explore new frontiers in physics.

Focusing on quantum phenomena like superposition instead of qualities based in the world of classical physics means we need even tinier readings than usual. Thats where quantum interferometry comes in, and its definitely more precise, but its also a work in progress. Over time, scientists are working to develop better and better ways to reduce noise (a quantum side effect) and other downsides to these methods.

In this research, scientists use an established paradigm called N00Nshort for a complex mathematical expressionwhere a cloud of n photons are all in the same state of superposition, and no non-superpositioned photons are allowed. A number of N00N states with two photons each are put into a prepared interferometer thats dozens of meters long on one side, using long fiber optic strands. Then, Earths rotation itself causes a measurable change within the photons.

It may seem silly to spend so much time and so many resources on clocking the speed of Earths rotationsomething we surely must know all about by now. A day is a certain length, and we know because we end up in the same place in about 24 hours ... right?

Mostly. But those facts reinforce what the problem is underneath. Earth is so enormous and steady that detecting tiny changes in its rotational state is actually very difficult.

To help, these researchers implemented helpful tools like a toggle that switches off how Earths rotation affected their interferometric setup, and they built their model to involve mapping the idea of rotation into a measurement that does not rotate like a calculus-based magic trick.

The results demonstrate not only the precision of their best-yet quantum interferometer, the researchers conclude, but also a milestone in our collective quest to marry general relativity and quantum mechanics into one unified theory. Indeed, being able to check the general, classical motion of the Earth against quantum phenomena could help move physics forward24 hours at a time.

Caroline Delbert is a writer, avid reader, and contributing editor at Pop Mech. She's also an enthusiast of just about everything. Her favorite topics include nuclear energy, cosmology, math of everyday things, and the philosophy of it all.

Read more from the original source:

Scientists Tracked Earth's Rotation With Frankly Unbelievable Precision - Popular Mechanics

Read More..

UNBC assistant professor makes groundbreaking discovery in the field of quantum physics – CKPGToday.ca

While magnons are known to occur naturally they do not want to appear next to each other and will repel as soon as possible, dissipating that energy. However when the researchers shone terahertz light waves to excite the spins in a material with the chemical composition of BaCo2V2O8 they created magnon pairs that were bound together with nowhere for that energy to dissipate.

Jobs

Year-over-year unemployment down substantially in Prince George

Transit

31 bus shelters to be upgraded

Animal Safety

Don't leave your animals in the car this summer

They remain next to each other for at least 18 picoseconds. So its a very short time. but its enough for the detection to occur, continues Bernier. And it was surprising that it exists because in a solid theres supposed to be a lot of channels for dissipation. So, it should have been really easy to for this state to dissipate. But it turns out that we can observe it which was the big breakthrough.

While uses for this breakthrough are theoretical it could have the potential to drastically increase privacy when telecommunicating.

Theres this race going on right now about trying to find ways to communicate using quantum effects, says Bernier. It is possible that these bound state could be used to alter transport in spin chains, which is one of the possible devices that could that could be used for quantum telecommunication.

For now, physicist will be on the hunt for more materials that contain these exotically bound objects.

Local news. Delivered. Free. Subscribe to our daily news wrap and get our top local stories delivered to your inbox every evening.

***

Email: sam.bennison@pattisonmedia.com

View post:

UNBC assistant professor makes groundbreaking discovery in the field of quantum physics - CKPGToday.ca

Read More..

New open-source software is greater than the sum of its parts – University of Waterloo

Accurate models of real-world scenarios are important for bringing theoretical and experimental research together in meaningful ways. Creating these realistic computer models, however, is a very large undertaking. Significant amounts of data, code, and expertise across a wide range of intricate areas are needed to create useful and comprehensive software.

Dr. Norbert Ltkenhaus, executive director of the Institute for Quantum Computing (IQC) and a professor in the University of Waterloos Department of Physics and Astronomy, alongside his research group, have spent the last several years developing accurate software models for research in quantum key distribution (QKD). QKD is a process for cryptography that harnesses fundamental principles of quantum mechanics to exchange secret keys, which can then be used to ensure secure communication.

Ltkenhaus and his research group recently released a modular, open-source software package on GitHub, which allows users to model realistic QKD protocols and calculate the generation rate for secure quantum keys using user-submitted variables for real-world scenarios.

Modelling and analyzing QKD setups require many different skills to come together. Our software framework allows experts in various areas like optimization theory, optical modelling and security analysis to bring their knowledge together, Ltkenhaus says. The open-source approach is designed to foster an interdisciplinary community from which all researchers will benefit.

Read more about this open-source QKD software package in the full story on Waterloo News.

See the original post here:

New open-source software is greater than the sum of its parts - University of Waterloo

Read More..

LLM Apps, Crucial Data Skills, Multi-AI Agent Systems, and Other July Must-Reads – Towards Data Science

Feeling inspired to write your first TDS post? Were always open to contributions from new authors.

If its already summer where you live, we hope youre making the most of the warm weather and (hopefully? maybe?) more relaxed daily rhythms. Learning never stops, of courseat least not for data scientistsso if your idea of a good time includes diving into new challenges and exploring cutting-edge tools and workflows, youre in for a treat.

Our July highlights, made up of the articles that created the biggest splash among our readers last month, cover a wide range of practical topicsand many of them are geared towards helping you raise your own bar and expand your skill set. Lets dive in!

Every month, were thrilled to see a fresh group of authors join TDS, each sharing their own unique voice, knowledge, and experience with our community. If youre looking for new writers to explore and follow, just browse the work of our latest additions, including Mengliu Zhao, Robbie Geoghegan, Alex Dremov, Torsten Walbaum, Jeremi Nuer, Jason Jia, Akchay Srivastava, Roman S, James Teo, Luis Fernando PREZ ARMAS, Ph.D., Lea Wu, W. Caden Hamrick, Jack Moore, Eddie Forson, Carsten Frommhold, Danila Morozovskii, Biman Chakraborty, Jean Meunier-Pion, Ken Kehoe, Robert Lohne, Pranav Jadhav, Cornellius Yudha Wijaya, Vito Rihaldijiran, Justin Laughlin, Yiit Ak, Teemu Sormunen, Lars Wiik, Rhea Goel, Ryan D'Cunha, Gonzalo Espinosa Duelo, Akila Somasundaram, Mel Richey, PhD, Loren Hinkson, Jonathan R. Williford, PhD, Daniel Low, Nicole Ren, Daniel Pollak, Stefan Todoran, Daniel Khoa Le, Avishek Biswas, Eyal Trabelsi, Ben Olney, Michael B Walker, Eleanor Hanna, and Magda Ntetsika.

Follow this link:

LLM Apps, Crucial Data Skills, Multi-AI Agent Systems, and Other July Must-Reads - Towards Data Science

Read More..

A data science roadmap for open science organizations engaged in early-stage drug discovery – Nature.com

Consistent data processing: a critical prelude to building AI models

The critical nature of precise storage, management, and dissemination of data in the realm of drug discovery is universally recognized. This is because the extraction of meaningful insights depends on the data being readily accessible, standardized, and maintained with the highest possible consistency. However, the implementation of good data practices can vary greatly and depends on the goals, culture, resources, and expertise of research organizations. A critical, yet sometimes underestimated, aspect is the initial engineering task of data preprocessing, which entails transforming raw assay data into a format suitable for downstream analysis. For instance, quantifying sequencing reads from DNA-encoded library screens into counts is required for the subsequent hit identification data science analysis step. Ensuring the correctness of this initial data processing step is imperative, but it may be given too little priority, potentially leading to inaccuracies in subsequent analyses. Standardization of raw data processing is an important step to enable subsequent machine learning studies of DEL data. Currently, this step is done by companies or organizations that generate and screen DEL libraries, and the respective protocols are reported if a study is published (see the Methods section in McCloskey et al. 18). Making data processing pipelines open source will help establish best practices to allow for scrutiny and revisions if necessary. While this foundational step is vital for harnessing data science, it is worth noting that it will not be the focus of this discussion.

In drug discovery, data science presents numerous opportunities to increase the efficiency and speed of the discovery process. Initially, data science facilitates the analysis of huge experimental data, e.g., allowing researchers to identify potential bioactive compounds in large screening data. Machine learning models can be trained on data from DEL or ASMS and, in turn, be used for hit expansion in extensive virtual screens. For example, a model trained to predict the read counts of a specific DEL screen can be used to identify molecules from other large compound libraries, which are likely to bind to the target protein under consideration18.

As the drug discovery process advances to compound optimization, data science can be used to analyse and predict the pharmacokinetic and dynamic properties of potential drug candidates. This includes model-based evaluation of absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiles. ADMET parameters are crucial in prioritizing and optimizing candidate molecules. Acknowledging their importance, the pharmaceutical industry has invested substantially in developing innovative assays and expanding testing capacities. Such initiatives have enabled the characterization of thousands of compounds through high-quality in-vitro ADMET assays, serving as a prime example of data curation in many pharmaceutical companies37. The knowledge derived from accumulated datasets has the potential to impact research beyond the projects where the data was originally produced. Computational teams utilize these data to understand the principles governing ADMET endpoints as well as to develop in-silico models for the prediction of ADMET properties. These models can help prioritize compound sets lacking undesired liabilities and thus guide researchers in their pursuit to identify the most promising novel drug candidates.

Major approaches in early drug discovery data science encompass classification, regression, or ranking models. They are, for example, employed in drug discovery to classify molecules as mutagenic, predict continuous outcomes such as the binding affinity to a target, and rank compounds in terms of their solubility. Incorporating prior domain knowledge can further enhance the predictive power of these models. Often, assays or endpoints that are correlated can be modelled together, even if they represent independent tasks. By doing so, the models can borrow statistical strength from each individual task, thereby improving overall performance compared to modelling them independently. For example, multitask learning models can predict multiple properties concurrently, as demonstrated by a multitask graph convolutional approach used for predicting physicochemical ADMET endpoints38.

When confronted with training data that have ambiguous labels, utilizing multiple-instance learning can be beneficial. Specifically, in the context of bioactivity models, this becomes relevant when multiple 3D conformations are considered, as the bioactive conformation is often unknown39. A prevalent challenge in applying data science for predictive modelling of chemical substances is choosing a suitable molecular representation. Different representations, such as Continuous Data-Driven Descriptor (CDDD)40 from SMILES strings, molecular fingerprints41 or 3D representations42, capture different facets of the molecular structure and properties43. It is vital to select an appropriate molecular representation as this determines how effectively the nuances of the chemical structures are captured. The choice of the molecular representation influences the prediction performance of various downstream tasks, making it a critical factor in AI-driven drug discovery, as discussed in detail in David et al.s43 review and practical guide on molecular representations in AI-driven drug discovery. Recent studies have found that simple k-nearest neighbours on molecular fingerprints can match or outperform much more complicated deep learning approaches on some compound potency prediction benchmarks44,45. On the other hand, McCloskey et al. 18 have discovered hits by training graph neural networks on data from DEL screens, which are not close to the training set using established molecular similarity calculations. Whether a simple molecular representation, infused with chemical knowledge, or a complex, data-driven deep learning representation is more suitable for the task at hand depends strongly on the training data and needs to be carefully evaluated on a case-by-case basis to obtain a fast and accurate model.

Sound strategies for splitting data into training and test sets are crucial to ensure robust model performance. These strategies include random splitting, which involves dividing the data into training and test sets at random, ensuring a diverse mix of data points in both sets. Temporal splitting arranges data chronologically, training the model on older data and testing it on more recent data, which is useful for predicting future trends. Compound cluster-wise splitting devides training and test sets into distinct chemical spaces. Employing these strategies is essential as inconsistencies between the distributions of training and test data can lead to unreliable model outputs, negatively impacting decision-making processes in drug discovery46.

The successful application of machine learning requires keeping their domain of applicability in mind at all stages. This includes using the techniques described in the previous section for data curation and model development. However, it is equally important to be able to estimate the reliability of a prediction made by an AI model. While generalization to unseen data is theoretically well understood for classic machine learning techniques, it is still an active area of research for deep learning. Neural networks can learn complex data representations through successive nonlinear transformations of the input. As a downside of this flexibility, these models are more sensitive to so-called adversarial examples, i.e., instances outside the domain of applicability that are seemingly close to the training data from the human perspective44. For this reason, deep learning models often fall short of providing reliable confidence estimates for their predictions. Several empirical techniques can be used to obtain uncertainty estimates: Neural network classifiers present a probability distribution indicative of prediction confidence, which is inadequately calibrated but can be adjusted on separate calibration data45. For regression tasks, techniques such as mixture density networks47 or Bayesian dropout48 can be employed to predict distributions instead of single-point estimates. For both classification and regression, the increased variance of a model ensemble indicates that the domain of applicability has been left49.

With the methods described in the previous paragraphs, we possess the necessary methodological stack to establish a data-driven feedback loop from experimental data, a crucial component for implementing active learning at scale. By leveraging predictive models that provide uncertainty estimates, we can create a dynamic and iterative data science process for the design-make-test-analyse (DMTA) cycle. For instance, these predictive models can be utilized to improve the potency of a compound by identifying and prioritizing molecules that are predicted to have high affinity yet are uncertain. Similarly, the models can be used to increase the solubility of a compound by selecting molecules that are likely to be more soluble, thus improving delivery and absorption. This process continuously refines predictions and prioritizes the most informative data points for subsequent experimental testing and retraining the predictive model, thereby enhancing the efficiency and effectiveness of drug discovery efforts. An important additional component is the strategy to pick molecules for subsequent experiments. By intelligently selecting the most informative samples, possibly those that the model is most uncertain about, the picking strategy ensures that each iteration contributes maximally to refining the model and improving predictions. For example, in the context of improving compound potency, the model might prioritize molecules that are predicted to have high potency but with a high degree of uncertainty. These strategies optimize the DMTA process by ensuring that each experimental cycle contributes to the refinement of the predictive model and the overall efficiency of the drug discovery process.

When applying the computational workflow depicted in Fig.3 on large compound libraries, scientists encounter a rather uncommon scenario for machine learning: usually, the training of deep neural networks incurs the highest computational cost since many iterations over large datasets are required, while comparatively few predictions will later be required from the trained model within a similar time frame. However, when inference is to be performed on a vast chemical space, we face the inverse situation. Assessing billions of molecules for their physicochemical parameters and bioactivity is an extremely costly procedure, potentially requiring thousands of graphics processing unit (GPU) hours. Therefore, not only predictive accuracy but also the computational cost of machine learning methods is an important aspect that should be considered when evaluating the practicality of a model.

Computational workflow for predicting molecular properties, starting with molecular structure encoding, followed by model selection and assessment, and concluding with the application of models to virtually screen libraries and rank these molecules for potential experimental validation. The process can be cyclical, allowing iterative refinement of models based on empirical data. ADMET: absorption, distribution, metabolism, and excretiontoxicity. ECFP: Extended Connectivity Fingerprints. CDDD: Continuous Data-Driven Descriptor, a type of molecular representation derived from SMILES strings. Entropy: Shannon entropy descriptors50,51.

Originally posted here:

A data science roadmap for open science organizations engaged in early-stage drug discovery - Nature.com

Read More..

Data science instruction comes of age – The Hechinger Report

This is an edition of our Future of Learning newsletter. Sign up today to get it delivered straight to your inbox.

Ive been reporting on data science education for two years now, and its become clear to me that whats missing is a national framework for teaching data skills and literacy, similar to the Common Core standards for math or the Next Generation Science Standards.

Data literacy is increasingly critical for many jobs in science, technology and beyond, and so far schools in 28 states offer some sort of data science course. But those classes vary widely in content and approach, in part because theres little agreement around what exactly data science education should look like.

Last week, there was finally some movement on this front a group of K-12 educators, students, higher ed officials and industry leaders presented initial findings on what they believe students should know about data by the time they graduate from high school.

Data Science 4 Everyone, an initiative based at the University of Chicago, assembled 11 focus groups that met over five months to debate what foundational knowledge on data and artificial intelligence students should acquire not only in dedicated data science classes but also in math, English, science and other subjects.

Among the groups proposals for what every graduating high schooler should be able to do:

On August 15, Data Science 4 Everyone plans to release a draft of its initial recommendations, and will be asking educators, parents and others across the country to vote on those ideas and give other feedback.

Here are a few key stories to bring you up to speed:

Data science under fire: What math do high schoolers really need?

Earlier this year, I reported on how a California school district created a data science course in 2020, to offer an alternative math course to students who might struggle in traditional junior and senior math courses such as Algebra II, Pre-Calculus and Calculus, or didnt plan to pursue science or math fields or attend a four-year college. California has been at the center of the debate on how much math, and what math, students need to know before high school graduation.

Eliminating advanced math tracks often prompts outrage. Some districts buck the trend

Hechinger contributor Steven Yoder wrote about how districts that try to detrack or stop sorting students by perceived ability often face parental pushback. But he identified a handful of districts that have forged ahead successfully with detracking.

PROOF POINTS: Stanfords Jo Boaler talks about her new book MATH-ish and takes on her critics

My colleague Jill Barshay spoke with Boaler, the controversial Stanford math education professor who has advocated for data science education, detracking and other strategies to change how math is taught. Jill writes that the academic fight over Boalers findings reflects wider weaknesses in education research.

Whats next: This summer and fall Im reporting on other math topics, including a program to get more Black and Hispanic students into and through Calculus, and efforts by some states to revise algebra instruction. Id love to hear your thoughts on these topics and other math ideas you think we should be writing about.

More on the Future of Learning

How did students pitch themselves to colleges after last years affirmative action ruling?, The Hechinger Report

PROOF POINTS: This is your brain. This is your brain on screens, The Hechinger Report

Budget would require districts to post plans to educate kids in emergencies, EdSource

Turmoil surrounds LAs new AI student chatbot as tech firm furloughs staff just 3 months after launch, The 74

Oklahoma education head discusses why hes mandating public schools teach the Bible, PBS

This story about data science standards was produced by The Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education.

The Hechinger Report provides in-depth, fact-based, unbiased reporting on education that is free to all readers. But that doesn't mean it's free to produce. Our work keeps educators and the public informed about pressing issues at schools and on campuses throughout the country. We tell the whole story, even when the details are inconvenient. Help us keep doing that.

Join us today.

Go here to read the rest:

Data science instruction comes of age - The Hechinger Report

Read More..

Forget Statistical Tests: A/B Testing Is All About Simulations – Towards Data Science

11 min read

Controlled experiments such as A/B tests are used heavily by companies.

However, many people are repelled by A/B testing due to the presence of intimidating statistical jargon including terms such as confidence, power, p-value, t-test, effect size, and so on.

In this article, I will show you that you dont need a Master in Statistics to understand A/B testing quite the opposite. In fact, simulations can replace most of those statistical artifacts that were necessary 100 years ago.

Not only this: I will also show you that the feasibility of an experiment can be measured using something that, unlike confidence and power, is understandable by anyone in the company: dollars.

Your website has a checkout page. The ML team has come out with a new recommendation model. They claim that, by embedding their recommendations into the checkout page, we can increase revenues by an astounding 5%.

See more here:

Forget Statistical Tests: A/B Testing Is All About Simulations - Towards Data Science

Read More..

Time Series Forecasting in the Age of GenAI: Make Gradient Boosting Behaves like LLMs – Towards Data Science

6 min read

The rise of Generative AI and Large Language Models (LLMs) has fascinated the entire world initializing a revolution in various fields. While the primary focus of this kind of technology has been on text sequences, further attention is now being given to expanding their capabilities to handle and process data formats beyond just text inputs.

Like in most AI areas, time series forecasting is also not immune to the advent of LLMs, but this may be a good deal for all. Time series modeling is known to be more like an art, where results are highly dependent on prior domain knowledge and adequate tuning. On the contrary, LLMs are appreciated for being task-agnostic, holding enormous potential in using their knowledge to solve variegated tasks coming from different domains. From the union of these two areas, the new frontier of time series forecasting models can be born which in the future will be able to achieve previously unthinkable results.

Read more from the original source:

Time Series Forecasting in the Age of GenAI: Make Gradient Boosting Behaves like LLMs - Towards Data Science

Read More..