Today’s seemingly insurmountable wall is symbolic reasoning, the capacity to manipulate symbols in the ways familiar from algebra or logic. As we learned as children, solving math problems involves a step-by-step manipulation of symbols according to strict rules (e.g., multiply the furthest right column, carry the extra value to the column to the left, etc.).” https://bit.ly/3mRhGii

“The second idea they drew on was the method of training the hypernetwork to make predictions for new candidate architectures. This requires two other neural networks. The first enables computations on the original candidate graph, resulting in updates to information associated with each node, and the second takes the updated nodes as input and predicts the parameters for the corresponding computational units of the candidate neural network.” https://bit.ly/3nanvaL

The laws of physics are symmetric through space. That means that the fundamental equations of gravity or electromagnetism or quantum mechanics apply equally throughout the entirety of the volume of the universe. They also work in any direction. So, a laboratory experiment that is rotated 90 degrees should produce the same results (all else being equal, of course).

But in a crystal, this gorgeous symmetry gets broken. The molecules of a crystal arrange themselves in a preferred direction, creating a repeating spatial structure. In the jargon of physicists, a crystal is a perfect example of “spontaneous symmetry breaking” — the fundamental laws of physics remain symmetric, but the arrangement of the molecules is not. https://bit.ly/3NoRw1d

Learning is an exotic process; until about a decade ago, brains were the only systems that did it well. It was the structure of the brain that loosely inspired computer scientists to design deep neural networks, now the most popular artificial learning models.

A deep neural network is a computer program that learns through practice. The network can be thought of as a grid: Layers of nodes called neurons, which store values, are connected to neurons in adjacent layers by lines, or “synapses.” Initially, these synapses are just random numbers known as “weights.” https://bit.ly/3c3NfmB

A growing number of experiments are implementing machine learning (ML) algorithms to aid in analyzing data, but these have the same limitations as the people they aim to help: They can’t directly access and learn from quantum information. But what if there were a quantum machine learning algorithm that could directly interact with this quantum data? https://bit.ly/3aOjYw2

Learn best practices to deploy language models - https://txt.cohere.ai/best-practices-for-deploying-language-models/

“Networks of nanoscale resistors that work in a similar way to nerve cells in the body could offer advantages over digital machine learning.

“Just as a human brain learns by remodelling the connections between millions of interconnected neurons, so too could machine learning models run on networks of these nanoresistors.” https://bit.ly/3PPrAO4

Transformers quickly became the front-runner for applications like word recognition that focus on analyzing and predicting text. It led to a wave of tools, like OpenAI’s Generative Pre-trained Transformer 3 (GPT-3), which trains on hundreds of billions of words and generates consistent new text to an unsettling degree. https://bit.ly/3QyzyeM

The basic idea of a quantum financial system is to allow for a broader application of cryptology and blockchain technology. Decentralized nodes would facilitate the process of completing transactions in the network, and quantum computing would be used for encryption purposes. https://bit.ly/3QAG0lz

Outlier Detection is very important in machine learning. We present a few techniques with examples - https://builtin.com/data-science/how-find-outliers-examples

PCPs (probabilistically checkable proof) have become some of the most important tools in theoretical computer science. Recently, they’ve even found their way into practical applications, such as in cryptocurrencies, where they are used for rolling up large batches of transactions into a smaller form that is easier to verify. https://bit.ly/3RjnJt9

Training a quantum neural network requires only a small amount of data, according to a new proof that upends previous assumptions stemming from classical computing’s huge appetite for data in machine learning, or artificial intelligence. The theorem has several direct applications, including more efficient compiling for quantum computers and distinguishing phases of matter for materials discovery. https://bit.ly/3TBcoGR

The term “entropy” is borrowed from physics, where entropy is a measure of disorder. A cloud has higher entropy than an ice cube, since a cloud allows for many more ways to arrange water molecules than a cube’s crystalline structure does. In an analogous way, a random message has a high Shannon entropy — there are so many possibilities for how its information can be arranged — whereas one that obeys a strict pattern has low entropy. https://bit.ly/3TQapyh

“While classic computing methods artificially create these hidden node models, they can be built naturally with qubits. The fundamental entanglement associated with backpropagation, a mathematical tool for improving the accuracy of predictions made by a machine learning model, can be computed much faster with qubits. This means training neural networks on quantum computers can be orders of magnitude faster.”

“Another space is nonstructured database searches, for which there is an ever-increasing set of problems that will exist as the internetworking of global computation creates massive amounts of data. While classical computers do an excellent job of searching through structured data, searches through unstructured data are much less efficient. Lov Grover, an Indian-American computer scientist, developed a quantum algorithm that can guarantee a dramatic speedup in searches. On small data sets, a speedup is not significant, but on large volumes of data, the practical speedups are significant.” https://bit.ly/3RTCvYl

A common approach for detecting outliers using descriptive statistics is the use of interquartile ranges (IQRs). This method works by analyzing the points that fall within a range specified by quartiles, where quartiles are four equally divided parts of the data. Although IQR works well for data containing a single shape or pattern, it is not able to distinguish different types of shapes or groups of data points within a data set.
Fortunately, clustering techniques address the limitations of IQR by effectively separating samples into different shapes. A commonly used clustering method for outlier detection is DBSCAN, which is an unsupervised clustering method that addresses many of the limitations of IQR….. https://bit.ly/3SNuILx

Tensor networks can help to alleviate the complexity of representing quantum states and operations by exploiting redundancies in the topological structure of the quantum circuit – instead of the vectors/ matrices. To translate a quantum circuit into a tensor network, each object, be it a state or an operation, is represented by a multi-dimensional array of complex numbers – a tensor. For a circuit, tensors are connected to other tensors according to the underlying quantum circuit. https://bit.ly/3fXlluR

Though hype may suggest otherwise, quantum computers aren’t all-powerful computational devices capable of solving intractable problems like the halting problem. Rather, they’re processors built on an architecture that we hope will allow them to do anything a classical computer can do, with extra capabilities or increased performance for some problems afforded by their quantum nature. https://bit.ly/3gm9eHM

“But unlike classical bits, qubits are extremely fragile. The physical objects that represent classical bits are made up of semiconductors. You can drop them on a table and they would still work fine. But if you so much as bumped against a table on which there is a functional qubit, it will break. Qubits are even affected by seemingly insignificant disturbances like stray electromagnetic waves, vibrations, temperature fluctuations and possibly cosmic rays.” https://tinyurl.com/5bzkfzdx

“Symplectic geometry is more complicated. Here, the answer depends on the ellipsoid’s “eccentricity,” a number that represents how elongated it is. A long, thin shape with a high eccentricity can be easily folded into a more compact shape, like a snake coiling up. When the eccentricity is low, things are less simple.” https://bit.ly/3DNw8B8

"In this article, it have discussed three waves of quantum machine learning, each harnessing a particular aspect of quantum computers and targeting particular problems. The first scrutinizes the power of quantum computers to work with high-dimensional data and speed-up algebra, but raises the caveat of input/output due to the quantum measurement rules. The second domain circumvents this problem by using a hybrid architecture, performing optimization on a classical computer while evaluating parameterized states on a quantum circuit, chosen based on a particular problem. Finally, the third domain is inspired by brain-like computation and uses the natural interaction and unitary dynamic of a given quantum system as a source for learning." shorturl.at/ackU5

“One of the key obstacles to accurate quantum simulations is noise—random errors in both the switching of the “gates” that perform quantum logic operations and in the reading of their output states. These errors accumulate and restrict the number of gate operations a computation can enact before the noise dominates. The researchers found that simulations with more than 300 gates were overwhelmed by noise. But the more complex the system, the more gates are needed.” shorturl.at/kpLPW