量子コンピュータ勉強中の機械学習エンジニアとして量子機械学習はぜひ学びたい。
おっ、量子Support Vector Machine(以下SVM)あるじゃん、ここから始めよう。
そう思ってひとまずQiitaを探したけど、まだこの辺は充実してない(2020/05時点)。
そもそもQiitaに限らず量子SVMの日本語の記事が少なく、あってもただ英語リソース翻訳したくらい。
実装して回してみた系の記事はほぼ皆無。
Starlink Will Transfer Data Close To 97% Of The Speed Of Light Says Musk
Space Exploration Technologies Corp.'s (SpaceX) chief Mr. Elon Musk has revealed that his company's Starlink satellite internet service will soon be capable of transferring data faster significantly closer to the speed of light. Mr. Musk's comments were made on the social media platform Twitter, where he shared details about new Starlink satellites, which SpaceX will launch in the next couple of months. These satellites will be equipped with lasers, which are a crucial feature that eliminates the need for Starlink to use ground stations to channel data to internet servers. According to the executive, the new satellites will launch soon and be active by the start of next year.
グーグル、アマゾンも参入。「何ができるか不明」でも巨額投資集まる「量子コンピューター」の現在位置
2021年7月27日、東京大学と日本IBMは、日本・アジア初となる商用量子コンピューター「IBM(R) Quantum System One」の稼働を開始したことを発表しました。
コンピューターは、現代社会を語る上で欠かせないものです。
ただ、「コンピューター」と一口に言っても、その種類はさまざまです。
私たちが日常生活の中で使うようなパソコンやスマートフォンもコンピューターですし、研究機関に設置されているようなとてつもない計算能力を持つ「スーパーコンピューター」もあります。
その中でも、ここ数年の間に耳にする機会が増えてきたのが、「量子力学」の原理を活用した「量子コンピューター」です。
Task: Maritime object detection and classification in global SAR scenes
Rapid detection of IUU fishing activity will enable interdiction and prosecution of offenders, mitigating the damages from IUU fishing. The rise of SAR satellite imagery offers an all-weather, day-and-night source to detect vessels that may otherwise elude fishing enforcement authorities.
We have constructed a large, multi-dimensional dataset of SAR satellite views and contextual data that are relevant for the tasks of maritime object detection and classification. Each xView3 data object represents a historical SAR satellite scene which has been preprocessed, co-registered, reprojected to Universal Transverse Mercator (UTM) coordinates, and used to match historical detections derived from AIS, VMS, automated, and manual (human) visual detections. For the xView3 Challenge, the prediction task is:
Ludwig is a toolbox that allows to train and test deep learning models without the need to write code.
A new data type-based approach to deep learning model design that makes the tool suited for many different applications.
Experienced users have deep control over model building and training, while newcomers will find it easy to use.
Easy to add new model architecture and new feature data-types.
パナソニック、AIプロセッサーを標準搭載したネットワークカメラ「i-PRO Sシリーズ」を9月発売
パナソニックi-PROセンシングソリューションズ株式会社は10日、ネットワークカメラの新しいスタンダードモデルとして「i-PRO Sシリーズ」を9月に発売し、パナソニック システムソリューションズ ジャパン株式会社を通じて販売すると発表した。
i-PRO Sシリーズは、ネットワークカメラの新しいスタンダードモデルとして、AIプロセッサーを標準で搭載し、カメラ内で顔や人、車両、二輪車を自動で識別できる。従来のネットワークカメラでは難しかった、映像の分析・解析といった高負荷のAI処理をカメラ内で行うことで、サーバー側で行っていたAI処理の負荷を分散する。
量子ゲート実機で簡単な回路をやってみて、どれぐらいの深さまで出来るのか試してみます。
使うのはIBMの ibmqx2 とします。(5量子ビット)
Algorithm: Semidefinite Programming
Speedup: Polynomial (with some exceptions)
Description: Given a list of m + 1 Hermitian n×n">n×nn×n matrices C,A1,A2,…,Am">C,A1,A2,…,AmC,A1,A2,…,Am and m numbers b1,…,bm">b1,…,bmb1,…,bm, the problem of semidefinite programming is to find the positive semidefinite n×n">n×nn×nmatrix X that maximizes tr(CX) subject to the constraints tr(AjX)≤bj">tr(AjX)≤bjtr(AjX)≤bj for j=1,2,…,m">j=1,2,…,mj=1,2,…,m. Semidefinite programming has many applications in operations research, combinatorial optimization, and quantum information, and it includes linear programming as a special case. Introduced in [313], and subsequently improved in [383, 425], quantum algorithms are now known that can approximately solve semidefinite programs to within ±ϵ">±ϵ±ϵ in time O(mlogm⋅poly(logn,r,ϵ−1))">O(m‾‾√logm⋅poly(logn,r,ϵ−1))O(mlogm⋅poly(logn,r,ϵ−1)), where r is the rank of the semidefinite program. This constitutes a quadratic speedup over the fastest classical algorithms when r is small compared to n. The quantum algorithm is based on amplitude amplification and quantum Gibbs sampling [121, 307]. In a model in which input is provided in the form of quantum states the quantum algorithm for semidefinite programming can achieve superpolynomial speedup, as discussed in [383], although recent dequantization results [421] delineate limitations on the context in which superpolynomial quantum speedup for semidefinite programs is possible.
机の上にあるのは、ノートPCと小型のモニターを組み合わせて作った箱のようなディスプレイ。ディスプレイには3Dキャラクターの踊る様子が映し出されているが、見る角度を変えるとまるで本当に机の上で踊っているかのよう。こんな様子を収めた動画がTwitterで話題になっている。
We give a quantum algorithm for solving semidefinite programs (SDPs). It has worst-case running time n12m12s2poly(log(n),log(m),R,r,1/δ), with n and s the dimension and row-sparsity of the input matrices, respectively, m the number of constraints, δ the accuracy of the solution, and R,r a upper bounds on the size of the optimal primal and dual solutions. This gives a square-root unconditional speed-up over any classical method for solving SDPs both in n and m. We prove the algorithm cannot be substantially improved (in terms of n and m) giving a Ω(n12+m12) quantum lower bound for solving semidefinite programs with constant s,R,r and δ.
The quantum algorithm is constructed by a combination of quantum Gibbs sampling and the multiplicative weight method. In particular it is based on a classical algorithm of Arora and Kale for approximately solving SDPs. We present a modification of their algorithm to eliminate the need for solving an inner linear program which may be of independent interest.
MLflow is an open source platform for managing the end-to-end machine learning lifecycle. It tackles four primary functions:
Tracking experiments to record and compare parameters and results (MLflow Tracking).
Packaging ML code in a reusable, reproducible form in order to share with other data scientists or transfer to production (MLflow Projects).
Managing and deploying models from a variety of ML libraries to a variety of model serving and inference platforms (MLflow Models).
Providing a central model store to collaboratively manage the full lifecycle of an MLflow Model, including model versioning, stage transitions, and annotations (MLflow Model Registry).
MLflow is library-agnostic. You can use it with any machine learning library, and in any programming language, since all functions are accessible through a REST API and CLI. For convenience, the project also includes a Python API, R API, and Java API.