Nobu_Portfolio
2020年1月6日月曜日
Best vegetarian multivitamin for IT professionals and CS students on a budget
Made in Canada, Webber Men's Most Complete Multi sold at Walmart regularly at 16.99 CAD on sale for 12.99 CAD right now.
The capsules have green powder in them, you get 90 capsules, take one per day, it just seems wayyy better for me than anything else available at that price point.
https://webbernaturals.com/en-ca/product/mens-most-complete-multi/
I swear to god that my coding skills in Python or any other language is way better on a vegetarian, vegan or no meat diet.
Meat just interferes with my chain of thought.
I finished search, basically maze solving, on AI for Robotics, Sebastien Thurn's course on Udacity and here's the .ipynb notebook:
https://github.com/nyck33/AI4RUdacity/blob/master/Lesson12Search.ipynb
I discovered a new usage in for loops where I decrement the counter in a DFS search where a particular branch leads to a dead end so it goes back to the previous level.
Next is A-star a heuristic algorithm that estimates the distance left to goal assuming there are no obstacles which helps in calculating an f = g +h where f is the term used to choose between nodes to expand, g is the distance travelled up to that point and h is the A-star estimate.
The capsules have green powder in them, you get 90 capsules, take one per day, it just seems wayyy better for me than anything else available at that price point.
https://webbernaturals.com/en-ca/product/mens-most-complete-multi/
I swear to god that my coding skills in Python or any other language is way better on a vegetarian, vegan or no meat diet.
Meat just interferes with my chain of thought.
I finished search, basically maze solving, on AI for Robotics, Sebastien Thurn's course on Udacity and here's the .ipynb notebook:
https://github.com/nyck33/AI4RUdacity/blob/master/Lesson12Search.ipynb
I discovered a new usage in for loops where I decrement the counter in a DFS search where a particular branch leads to a dead end so it goes back to the previous level.
Next is A-star a heuristic algorithm that estimates the distance left to goal assuming there are no obstacles which helps in calculating an f = g +h where f is the term used to choose between nodes to expand, g is the distance travelled up to that point and h is the A-star estimate.
2019年4月5日金曜日
Solved MtnCarContinuous using Udacity project code as boilerplate
episode: 0 score: -41.59034754863541 mean: -41.59 std: 0.0
episode: 1 score: -75.87242108451743 mean: -58.73 std: 17.14
episode: 2 score: 32.01844640263133 mean: -28.48 std: 45.01
episode: 3 score: 132.904319261567 mean: 11.86 std: 80.02
episode: 4 score: 125.81198290946529 mean: 34.65 std: 84.85
episode: 5 score: 84.26163480413017 mean: 42.92 std: 79.63
episode: 6 score: 126.89684490110164 mean: 54.92 std: 79.37
episode: 7 score: 139.18190524840517 mean: 65.45 std: 79.3
episode: 8 score: 100.24481691450521 mean: 69.32 std: 75.56
episode: 9 score: 165.80286734425076 mean: 78.97 std: 77.31
episode: 10 score: 109.29507292352991 mean: 94.05 std: 66.24
episode: 11 score: 209.07900825070152 mean: 122.55 std: 44.84
episode: 1 score: -75.87242108451743 mean: -58.73 std: 17.14
episode: 2 score: 32.01844640263133 mean: -28.48 std: 45.01
episode: 3 score: 132.904319261567 mean: 11.86 std: 80.02
episode: 4 score: 125.81198290946529 mean: 34.65 std: 84.85
episode: 5 score: 84.26163480413017 mean: 42.92 std: 79.63
episode: 6 score: 126.89684490110164 mean: 54.92 std: 79.37
episode: 7 score: 139.18190524840517 mean: 65.45 std: 79.3
episode: 8 score: 100.24481691450521 mean: 69.32 std: 75.56
episode: 9 score: 165.80286734425076 mean: 78.97 std: 77.31
episode: 10 score: 109.29507292352991 mean: 94.05 std: 66.24
episode: 11 score: 209.07900825070152 mean: 122.55 std: 44.84
Solved means getting over 90 for reward. Scores are well over 90 since I used a modified reward function.
2019年2月9日土曜日
2019年2月4日月曜日
DQN Cartppole-V1 にてTensorboardを使って学習をグラフ化しました
上記のグラフのLossの下りと同時にスコアは改善されてます:
episode: 0 score: 10.0 memory length: 11 epsilon: 1
episode: 5 score: 76.0 memory length: 154 epsilon: 1
episode: 10 score: 16.0 memory length: 278 epsilon: 1
episode: 15 score: 17.0 memory length: 364 epsilon: 1
episode: 20 score: 17.0 memory length: 491 epsilon: 1
episode: 25 score: 32.0 memory length: 596 epsilon: 1
episode: 30 score: 12.0 memory length: 657 epsilon: 1
episode: 35 score: 20.0 memory length: 742 epsilon: 1
episode: 40 score: 16.0 memory length: 846 epsilon: 1
episode: 45 score: 19.0 memory length: 954 epsilon: 1
self.epsilon 0.999
episode: 50 score: 19.0 memory length: 1126 epsilon: 0.8806777104745716
episode: 55 score: 9.0 memory length: 1238 epsilon: 0.7873207291459607
episode: 60 score: 17.0 memory length: 1389 epsilon: 0.6769247732130653
episode: 65 score: 14.0 memory length: 1545 epsilon: 0.5791040088995179
episode: 70 score: 65.0 memory length: 1821 epsilon: 0.4393709323780249
episode: 75 score: 60.0 memory length: 2000 epsilon: 0.3104958044435009
episode: 80 score: 119.0 memory length: 2000 epsilon: 0.20622457658762192
episode: 85 score: 201.0 memory length: 2000 epsilon: 0.07402874109670564
episode: 90 score: 239.0 memory length: 2000 epsilon: 0.02485123742451863
episode: 95 score: 284.0 memory length: 2000 epsilon: 0.009998671593271896
episode: 100 score: 202.0 memory length: 2000 epsilon: 0.009998671593271896
episode: 105 score: 292.0 memory length: 2000 epsilon: 0.009998671593271896
Tensorboardの使い方はこちらが分かりやすいです:https://stackoverflow.com/questions/42112260/how-do-i-use-the-tensorboard-callback-of-keras
以上でした。
マレーシアのBPOにてフロントエンドエンジニアを募集
マレーシアの首都KLのBPO、Teledirect社がフロントエンドエンジニアを募集中です。
Google Tag Manager, Google Ads, Google Merchant Centerの専門知識が得られます。
JavascriptにてTriggerをHTML内に書き込み(EventListner)、Eventが発生した際いにDataLayer内のVariableにデータを移してそこからGTM Container経由にてGoogle Analytics等に情報を送信します。面接の前に少し勉強する必要性がありますが、主にJavascript及びHTMLのスキルが評価されます。
給料は10000リンギット(267,227.87 円)から14000リンギット(374,119.02 円)ですが、物価が大体日本の3分の1(電化製品などの輸入品以外)なので悪くないと思います。
興味のある方はメールにてご連絡お願いします。
NyckimstrongandfreeアットマークGmail.com
Google Tag Manager, Google Ads, Google Merchant Centerの専門知識が得られます。
JavascriptにてTriggerをHTML内に書き込み(EventListner)、Eventが発生した際いにDataLayer内のVariableにデータを移してそこからGTM Container経由にてGoogle Analytics等に情報を送信します。面接の前に少し勉強する必要性がありますが、主にJavascript及びHTMLのスキルが評価されます。
給料は10000リンギット(267,227.87 円)から14000リンギット(374,119.02 円)ですが、物価が大体日本の3分の1(電化製品などの輸入品以外)なので悪くないと思います。
興味のある方はメールにてご連絡お願いします。
NyckimstrongandfreeアットマークGmail.com
登録:
投稿 (Atom)
My Github repo
In case anyone is interested: https://github.com/nyck33
-
Deep Q Networkを改善してCartpole-V1を140エピソードで攻略しました(最寄りのスコア10個の平均値が500を達成した時点で終了)。 ソースコードは借りたものですが、Experience Replay用のBufferが十分に埋まるまでEpsilon値を削...
-
https://www.economist.com/technology-quarterly/2015/03/05/silicon-valley-gets-a-taste-for-food
-
まずはRNNの復習を終わったところでRNN及びLSTMをDeep Reinforcement Learningで使えるかどうか調べたら、このようなリサーチペーパーがありました: DRQN?! . They have, however, struggled with learn...