投稿時間:2022-10-21 17:28:15 RSSフィード2022-10-21 17:00 分まとめ(29件)

カテゴリー等	サイト名等	記事タイトル・トレンドワード等	リンクURL	頻出ワード・要約等/検索ボリューム	登録日
フリーソフト	新着ソフトレビュー - Vector	ユーザーの声から新機能を搭載したインターネットラジオ再生録音ソフト「ネットラジオ録音 X2」	https://www.vector.co.jp/magazine/softnews/221021/n2210211.html?ref=rss	録音	2022-10-21 17:00:00
IT	ITmedia 総合記事一覧	[ITmedia ビジネスオンライン] 長引くガソリン価格高騰　「高い！」と感じるレギュラー価格」はいくら？	https://www.itmedia.co.jp/business/articles/2210/21/news140.html	itmedia	2022-10-21 16:39:00
IT	ITmedia 総合記事一覧	[ITmedia PC USER] 第13世代Coreプロセッサ採用BTOデスクトップPCが各社から発売	https://www.itmedia.co.jp/pcuser/articles/2210/21/news143.html	itmediapcuser	2022-10-21 16:10:00
AWS	AWS Japan Blog	【開催報告】「AWS 秋の Amazon EC2 Deep Dive 祭り 2022」セミナー	https://aws.amazon.com/jp/blogs/news/event-report-wwso-compute-ec2-20221013/	amazonecdeep	2022-10-21 07:35:52
AWS	AWS - Webinar Channel	Evaluating an Identity Verification Solution	https://www.youtube.com/watch?v=bH4y72-pBhk	identity	2022-10-21 07:03:35
AWS	AWS - Japan	Amazon SageMaker 推論 Part3（後編）もう悩まない！機械学習モデルのデプロイパターンと戦略【ML-Dark-05b】【AWS Black Belt】	https://www.youtube.com/watch?v=7pScGkPped8	AmazonSageMaker推論Part後編もう悩まない機械学習モデルのデプロイパターンと戦略【MLDarkb】【AWSBlackBelt】AmazonSageMakerを用いて機械学習モデルをデプロイする際に選べるデプロイオプションとコスト削減の機能についてまとめ、それぞれをどう選択するか、判断基準を解説します。	2022-10-21 07:50:06
AWS	lambdaタグが付けられた新着投稿 - Qiita	NAT Gawatewayを使ったLambdaのIPアドレスの固定化方法	https://qiita.com/Shinkijigyo_no_Hitsuji/items/06b25f6f7052b852cade	lambda	2022-10-21 16:08:01
python	Pythonタグが付けられた新着投稿 - Qiita	【MubertAI】テキストを音楽に変えるAIでBeatlesとQueenのコラボが実現	https://qiita.com/DeepRecommend/items/2b031830acf1ab3b561e	mubertai	2022-10-21 16:46:16
js	JavaScriptタグが付けられた新着投稿 - Qiita	🔰JS初心者が作るGoogle extension V3 ②	https://qiita.com/mmaumtjgj/items/ef7bf1116988bfeedd84	googleextensionv	2022-10-21 16:44:59
AWS	AWSタグが付けられた新着投稿 - Qiita	NAT Gawatewayを使ったLambdaのIPアドレスの固定化方法	https://qiita.com/Shinkijigyo_no_Hitsuji/items/06b25f6f7052b852cade	lambda	2022-10-21 16:08:01
golang	Goタグが付けられた新着投稿 - Qiita	Go言語入門 学習メモ 9 Stringerインタフェース、Errorインタフェース	https://qiita.com/fsd-osw/items/8d21babcd2f3b644b976	error	2022-10-21 16:09:02
技術ブログ	Developers.IO	オンデマンドキャパシティ予約で利用可能な数量が確保されているのにも関わらず Insufficient Instance Capacity エラーが発生したときの対処方法	https://dev.classmethod.jp/articles/tsnote-ec2-on-demand-capacity-reservation-insufficient-instance-capacity-az-attributes-01/	instancecapacity	2022-10-21 07:37:57
海外TECH	DEV Community	How to Set up Facebook Pixel on Shopify	https://dev.to/gloriamaldonado/how-to-set-up-facebook-pixel-on-shopify-5aaf	How to Set up Facebook Pixel on Shopify What is Facebook Pixel Facebook Pixel can be instrumental for you to better understand your target audience and grow your business as a Shopify store owner It offers valuable insights you can capitalize on to create much more effective Facebook ads targeting the right audience Once you allow the pixel to track the user actions amp behavior on your Shopify store coming through a Facebook ad you can ramp up your conversions by Monitoring when a customer takes action following your Facebook adFocusing on retargeting particular customers thanks to the Customer Audience featureAdvertising for people who are more likely to take action and convertSeeing which products attract more interest by analyzing every action on your website Measuring cross device conversionsRead on and get yourself a detailed amp step by step guide on how to make Facebook Pixel work on your Shopify store How Does Facebook Pixel Work with Shopify Once you integrate Facebook Pixel functions as a tracking pixel that lets you analyze how effective your ad campaigns are It operates on the browser side by transferring the user behavior amp data to Facebook when users visit your website Browser side also known as client side tracking is the typical tracking method that is preferred by nearly all websites When you opt for this sort of tracking it means that almost every user action takes place in the user s browser On server side tracking though you get a specific Google Cloud server that receives all the requests from your website The pixels related to both marketing and tracking are rendered through this server rather than the browser which provides better results The native Facebook Pixel integration works well for Shopify stores although you might want to set up the Facebook Conversion API to get even better results because of some inaccuracy problems caused by browser side tracking How to Set up Facebook Pixel on Shopify Here I share a step process so you can get Facebook Pixel up and running on your Shopify store Before you start you should make sure to ​​Install the official Facebook Channel App on your Shopify storeHave the Facebook Business Manager with admin rights thereNow you can move on with the steps Step Navigate to Shopify Admin gt Online Store gt Preferences gt Setup FacebookStep Authorize your FB Business Manager account and choose the Pixel later on Step You should choose “Maximum here in order to make sure all the data is shared and then confirm Step Go to FB Pixel Settings gt Automated Advanced Matching and set all the fields from Email to External ID ON to finalize the process Congratulations Now that you ve completed the Facebook Pixel Shopify setup you ve got yourself an effective amp free tool that can help to show your ads to the right people which can let you increase your conversions at the end of the day You can track your conversions understand your audience by using a vast variety of metrics and level up your remarketing campaigns	2022-10-21 07:39:39
海外TECH	DEV Community	AI learning how to land on the moon	https://dev.to/ashwinscode/teaching-an-ai-to-land-on-the-moon-5138	AI learning how to land on the moonWelcome everyone to this post where I teach an AI how to land on the moon Of course I am not talking about the actual moon although I wish I was however I am talking about a simulated environment instead OpenAI s Lunar Lander Gym environment Reinforcement LearningI recently came across an article about DeepMind s AlphaTensor You can read about it hereAlphaTensor learnt how to multiply two matrices together in an extremely efficient way managing to complete multiplications in fewer steps than Strassen s algorithm the previous best algorithm Reading this article inspired me to read into an area of machine learning I did not know much about reinforcement learning DeepMind has built several other very impressive AIs which were all trained using reinforcement learning algorithms What is reinforcement learning Reinforcement learning concerns learning the best actions to take in certain situations in an environment in order to achieve a certain goal For example in a game of chess RL algorithms would learn what piece to move and where to move it given the state of the game board and the goal of winning the game RL models learn purely from their interactions in an environment They are given no training dataset with what are the best actions to take in a given situation They learn everything from experience How do they learn An agent is anything that interacts with an environment by observing it and taking actions based on those observations For each action the agent takes in an environment the agent is given a reward which indicates how good that action was given the observation and the aim of the agent in the environment RL algorithms improve the performance of these agents essentially through trial and error They initially perform random actions and see what rewards they get from them They then can develop a policy over time based on these action reward pairs A policy describes the best actions to take in given situations with the goal of the environment kept in mind There are several algorithms available for developing a policy For this project I used a DQN Deep Q Network to develop a policy Q Values and DQNsDQNs are neural networks that take in the state of the environment as input and output the q values for each possible action that the agent can take It describes the agent s policy There is no fixed model architecture for DQNs It varies from environment to environment For example an agent playing an Atari game might observe the environment through a picture of the game In the case it would be best to use convolutional layers as part of the DQN architecture Using a simple feedforward neural network would work fine with other environments such as the lunar lander environment in OpenAI s Gym Q values measure the expected future rewards when taking that action assuming that the same policy is followed in the future When an agent is following a policy it takes the action with the greatest q value at the current state it s in The way DQNs are trained to determine accurate q values will be explained as I go through the code DQNs are used when the action space in an environment is discrete i e there is a finite number of possible actions an agent can take in an environment For example if an environment involved driving a car its action space would be considered discrete if the only actions allowed were to drive forward turn left and right degrees Its action space would be continuous if actions involved specifying the angle of the steering wheel and the speed to travel at This is because these can take an infinite number of values and therefore there an infinite number of actions Problem and CodeAs I mentioned earlier I am going to train an agent within OpenAI s Lunar Lander Gym environment The aim of the agent in this environment is to land the lander on its legs between the two flags The agent can take actions Do nothingFire left engineFire right engineFire main engineAn observation taken from this environment is an dimensional vector containing X coordinateY coordinateX velocityY velocityAngle of the landerAngular velocity Booleans describing whether each leg is in contact with the ground or notThe agent is rewarded as follows points for crashing points for coming to rest points for each leg in contact with the ground points for each frame firing main engine points for each frame firing a side engine points for moving from top to the landing padThe agent is considered to have solved the environment if it has collected at least points in an episode episode series of steps frames that occur until some criteria for the environment to reset has been met episode termination Episodes terminate if the lander crashesthe lander goes out of view horizontally Training a DQNAn agent interacts with an environment by taking actions in it For each step in the environment the agent records the following into a replay memory The observation it took of the environmentThe action it took from this observationThe reward it gainedThe new observation of the environmentWhether the episode has terminated or notThe agent has an initial exploration rate which determines how often it should take random moves instead of taking actions from the DQN s policy This is so that all actions can be explored during the training phase and therefore allow the algorithm to see which actions would be best in certain situations The DQN s initial policy is random so having the agent follow it all the time would mean the DQN would struggle to train to develop a strong policy since different actions haven t been explored for the same states The exploration rate decreases by an appropriate rate after each episode By the time the exploration rate reaches the agent will follow the DQN s policy only By this time the DQN should have produced a strong policy Q s a r γQ s′ a′ Q s a r γQ s a Q s a r γQ s′ a′ QQ Q is the policy function Takes in the environment state and an action as input and returns the q value for that action ss s is the current state aa a is a possible action rr r the reward for taking action aa a at state ss s s′s s′ the state of the environment after taking action aa a at state ss s γγγ is the discount rate It is a constant determined by a human measuring how important future actions are in the environment For every n steps in the environment a random batch is taken from the replay memory The DQN then predicts the q values at each state in the batch Q s a Q s a Q s a and the q values at each of the new states so that the best action at that state can be obtained Q s′ a′ Q s a Q s′ a′ For each item in the batch the calculated Q s a Q s a Q s a Q s′ a′ Q s a Q s′ a′ and reward values are substituted into the equation above This should calculate a slightly better Q s a Q s a Q s a value for this batch item The DQN is then trained with the batch observations as input and the newly calculated Q s a Q s a Q s a values as output Note calculating Q s a Q s a Q s a and the Q s′ a′ Q s a Q s′ a′ values are done by separate networks the policy and target network They are initialised with the same weights The policy network is the main network that is trained The target network isn t trained however the policy network s weights are copied to for some every m steps in the environment This is done so that the training process becomes stable If one network was used to predict both Q s a Q s a Q s a and Q s′ a′ Q s a Q s′ a′ and trained the network would end up be chasing a forever moving target leading to poor results The use of the target network to calculate Q s′ a′ Q s a Q s′ a′ means that the policy network has a still target to aim at for a while before the target changes instead of the target changing ever step in the environment This is repeated for a specified number of steps Over time the policy should become stronger class DQN def init self action n model self action n action n self policy model action n self target model action n self replay self max replay size self weights initialised False def play episode self env epsilon max timesteps obs env reset rewards steps for in range max timesteps rand np random uniform taking a random action or the action described by the DQN policy if rand lt epsilon action env action space sample else actions self policy np array obs astype float numpy action np argmax actions if not self weights initialised self target set weights self policy get weights self weights initialised True new obs reward done env step action if len self replay gt self max replay size self replay self replay len self replay self max replay size save data into replay memory for training self replay append obs action reward new obs done count rewards and steps so that we can see some information during training rewards reward obs new obs steps yield steps rewards if done env close break def learn self env timesteps train every update target every show every episode batch size discount min epsilon min reward max episode timesteps episodes epsilon exploration rate decay np e np log min epsilon timesteps how much the exploration rate should reduce each episode steps episode list rewards list while steps lt timesteps for ep len rewards in self play episode env epsilon max episode timesteps epsilon decay steps if steps train every and len self replay gt batch size taking random batch from replay memory batch random sample self replay batch size obs np array o for o in batch new obs np array o for o in batch calculating the Q s a values curr qs self policy obs numpy calculating q values of the future new observations to obtain Q s a future qs self target new obs numpy for row in range len batch action batch row reward batch row done batch row if not done Q s a reward Q s a curr qs row action reward discount np max future qs row else if the environment is completed there are no future actions so Q s a reward only curr qs row action reward fitting DQN to newly calculated Q s a values self policy fit obs curr qs batch size batch size verbose if steps update target every and len self replay gt batch size updating target model self target set weights self policy get weights episodes showing some training data if episodes show every episode print epsiode episodes print explore rate epsilon print episode reward rewards print episode length ep len print timesteps done steps if rewards gt min reward self policy save f policy model rewards episode list append episodes rewards list append rewards self policy save policy model final plt plot episode list rewards list plt show DQN pyNow that training is out the way here is the code for the whole DQN py file import numpy as npimport tensorflow as tfimport randomfrom matplotlib import pyplot as pltdef build dense policy nn def f action n model tf keras models Sequential tf keras layers Dense activation relu tf keras layers Dense activation relu tf keras layers Dense activation relu tf keras layers Dense activation relu tf keras layers Dense action n activation linear model compile loss tf keras losses MeanSquaredError optimizer tf keras optimizers Adam return model return fclass DQN def init self action n model self action n action n self policy model action n self target model action n self replay self max replay size self weights initialised False def play episode self env epsilon max timesteps obs env reset rewards steps for in range max timesteps rand np random uniform if rand lt epsilon action env action space sample else actions self policy np array obs astype float numpy action np argmax actions if not self weights initialised self target set weights self policy get weights self weights initialised True new obs reward done env step action if len self replay gt self max replay size self replay self replay len self replay self max replay size self replay append obs action reward new obs done rewards reward obs new obs steps yield steps rewards if done env close break def learn self env timesteps train every update target every show every episode batch size discount min epsilon min reward max episode timesteps episodes epsilon decay np e np log min epsilon timesteps steps episode list rewards list while steps lt timesteps for ep len rewards in self play episode env epsilon max episode timesteps epsilon decay steps if steps train every and len self replay gt batch size batch random sample self replay batch size obs np array o for o in batch new obs np array o for o in batch curr qs self policy obs numpy future qs self target new obs numpy for row in range len batch action batch row reward batch row done batch row if not done curr qs row action reward discount np max future qs row else curr qs row action reward self policy fit obs curr qs batch size batch size verbose if steps update target every and len self replay gt batch size self target set weights self policy get weights episodes if episodes show every episode print epsiode episodes print explore rate epsilon print episode reward rewards print episode length ep len print timesteps done steps if rewards gt min reward self policy save f policy model rewards episode list append episodes rewards list append rewards self policy save policy model final plt plot episode list rewards list plt show def play self env for in range obs env reset done False while not done actions self policy np array obs astype float numpy action np argmax actions obs done env step action env render def load self path m tf keras models load model path self policy mplay is the method that shows the agent in action load loads a saved DQN into the class Testing it out import gymfrom dqn import env gym make LunarLander v dqn DQN build dense policy nn dqn play env dqn learn env dqn play env Before training the agent plays like this During training we can see how it s going epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done epsiode explore rate episode reward episode length timesteps done and this episode vs reward graph You might expect there to be a clearer trend showing the reward increasing over time However due to the exploration rate this trend is distorted As the episodes go on the exploration rate decreases so the expected trend of rewards increasing over time becomes slightly more apparent Here is how the agent performs at the end of the training process It could still do with a smoother landing but I think this is a good performance nonetheless Maybe you could try this out yourself and tweak some of the training parameters and see what results they yield	2022-10-21 07:26:17
医療系	医療介護 CBnews	副反応疑い報告基準に「熱性けいれん」追加へ-コロナワクチン接種後7日以内の発症	https://www.cbnews.jp/news/entry/20221021164246	予防接種法	2022-10-21 16:55:00
金融	JPX マーケットニュース	[JPX総研]株価指数算出上の取扱いについて（GFA）	https://www.jpx.co.jp/news/6030/20221021-01.html	gfajpx	2022-10-21 16:20:00
海外ニュース	Japan Times latest articles	Cabinet OKs bill to rebalance Lower House electoral districts	https://www.japantimes.co.jp/news/2022/10/21/national/politics-diplomacy/lower-house-bill/	Cabinet OKs bill to rebalance Lower House electoral districtsThe revision to the Public Offices Election Act would affect single seat districts in prefectures with an eye to narrowing vote disparity	2022-10-21 16:18:39
海外ニュース	Japan Times latest articles	Tokyo Olympics bribery scandal: Investigation ensnares stuffed-toy maker and ad firms	https://www.japantimes.co.jp/news/2022/10/21/national/tokyo-olympics-bribery-scandal-explainer/	Tokyo Olympics bribery scandal Investigation ensnares stuffed toy maker and ad firmsThe case has grown to encompass ADK Holdings ーJapan s third largest advertising firm ーand stuffed toy maker Sun Arrow among others	2022-10-21 16:16:40
海外ニュース	Japan Times latest articles	Kei Komuro, husband of former Princess Mako, passes New York bar exam	https://www.japantimes.co.jp/news/2022/10/21/national/kei-komuro-new-york-bar-exam-pass/	Kei Komuro husband of former Princess Mako passes New York bar examThe result which came at the third attempt will likely bring relief to a couple dogged by intense criticism and scrutiny from the Japanese public	2022-10-21 16:10:35
ニュース	BBC News - Home	Keir Starmer leads calls for immediate general election	https://www.bbc.co.uk/news/uk-politics-63328852?at_medium=RSS&at_campaign=KARANGA	truss	2022-10-21 07:52:21
ニュース	BBC News - Home	UK economy hit as people shop less than pre-Covid	https://www.bbc.co.uk/news/business-63340725?at_medium=RSS&at_campaign=KARANGA	statistics	2022-10-21 07:54:56
ニュース	BBC News - Home	T20 World Cup: Ireland beat West Indies to advance in Hobart	https://www.bbc.co.uk/sport/cricket/63311585?at_medium=RSS&at_campaign=KARANGA	hobart	2022-10-21 07:31:47
ニュース	BBC News - Home	T20 World Cup: Ireland's Paul Stirling hits 50 against the West Indies	https://www.bbc.co.uk/sport/av/cricket/63341809?at_medium=RSS&at_campaign=KARANGA	T World Cup Ireland x s Paul Stirling hits against the West IndiesWatch the best shots from Paul Stirling s half century as Ireland beat the West Indies by nine wickets to qualify for the Super stage at the T World Cup in Hobart Australia	2022-10-21 07:30:31
ビジネス	不景気.com	衣料品製造「ミックコーポレーション」が破産へ、負債10億円 - 不景気com	https://www.fukeiki.com/2022/10/mic-corp.html	株式会社	2022-10-21 07:20:07
北海道	北海道新聞	室蘭の日鉄で作業員２人死亡　コークス炉でメンテナンス中	https://www.hokkaido-np.co.jp/article/748775/	室蘭市仲町	2022-10-21 16:03:47
ビジネス	東洋経済オンライン	｢主治医に遠慮不要｣セカンドオピニオンの受け方 どんな時に必要か｢ポイント3つ｣を医師が解説 | 読書 | 東洋経済オンライン	https://toyokeizai.net/articles/-/623156?utm_source=rss&utm_medium=http&utm_campaign=link_back	東洋経済オンライン	2022-10-21 16:30:00
ビジネス	プレジデントオンライン	大学受験は｢人生をやり直す方法｣のひとつ…私が不登校やひきこもりの人に学び直しを勧めるワケ - ｢高校まで｣と｢大学から｣はまったく違う	https://president.jp/articles/-/62757	大学受験	2022-10-21 17:00:00
IT	週刊アスキー	4年ぶりの区民まつりを大いに楽しもう！　「令和4年度ほどがや区民まつり」10月29日開催	https://weekly.ascii.jp/elem/000/004/109/4109946/	保土ケ谷	2022-10-21 16:30:00
IT	週刊アスキー	SwitchにRE ENGINE制「バイオハザード」が続々登場！まずは『バイオハザード ヴィレッジ クラウド』が10月28日に発売	https://weekly.ascii.jp/elem/000/004/109/4109950/	nintendo	2022-10-21 16:10:00