JPWO2020208444A5 - - Google Patents

Download PDF

Info

Publication number
JPWO2020208444A5
JPWO2020208444A5 JP2021558964A JP2021558964A JPWO2020208444A5 JP WO2020208444 A5 JPWO2020208444 A5 JP WO2020208444A5 JP 2021558964 A JP2021558964 A JP 2021558964A JP 2021558964 A JP2021558964 A JP 2021558964A JP WO2020208444 A5 JPWO2020208444 A5 JP WO2020208444A5
Authority
JP
Japan
Prior art keywords
computer
mlm
fairness
implemented method
initial
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP2021558964A
Other languages
Japanese (ja)
Other versions
JP2022527536A (en
Publication date
Priority claimed from US16/377,727 external-priority patent/US20200320428A1/en
Application filed filed Critical
Publication of JP2022527536A publication Critical patent/JP2022527536A/en
Publication of JPWO2020208444A5 publication Critical patent/JPWO2020208444A5/ja
Pending legal-status Critical Current

Links

Claims (10)

コンピュータ実施方法であって、
複数のパラメータ値と、複数のハイパーパラメータ値と、セグメント化された関連するサブグループに関する公平性を反映する初期公平性値とを含む機械学習モデル（ＭＬＭ）の初期バージョンを受信するステップと、
前記ＭＬＭの暫定バージョンを作成するために、前記ＭＬＭの前記初期バージョンの前記パラメータ値のうちの少なくとも一部または前記ハイパーパラメータ値のうちの少なくとも一部あるいはその両方を調整するステップと、
前記ＭＬＭの前記暫定バージョンの公平性値を決定するステップであって、
複数の公平性関連目標値と前記複数の公平性関連目標値を反映する報酬関数とを定義する強化学習メタ・モデル（ＲＬＭＭ）を受信することと、
前記ＭＬＭの前記暫定バージョンを動作させることと、
前記ＭＬＭの前記暫定バージョンの前記動作中に、前記ＲＬＭＭによって、前記報酬関数に基づいて報酬値を計算することと、
前記報酬値に基づいて前記ＭＬＭの前記暫定バージョンの暫定公平性値を決定することとを含む動作によって、決定するステップと、
前記暫定公平性値が前記初期公平性値よりも大きいことを決定するステップと、
前記暫定公平性値が前記初期公平性値よりも大きいという前記決定に応答して、前記ＭＬＭの前記初期バージョンを前記ＭＬＭの前記暫定バージョンに置き換え、前記初期公平性値を前記暫定公平性値に置き換えるステップとを含む、コンピュータ実施方法。 A computer-implemented method comprising:
receiving an initial version of a machine learning model (MLM) including a plurality of parameter values, a plurality of hyperparameter values, and an initial fairness value reflecting fairness with respect to the segmented relevant subgroups;
adjusting at least some of the parameter values and/or at least some of the hyperparameter values of the initial version of the MLM to create an interim version of the MLM;
determining a fairness value for the interim version of the MLM, comprising:
receiving a reinforcement learning meta-model (RLMM) defining a plurality of fairness-related goals and a reward function reflecting the plurality of fairness-related goals;
running the interim version of the MLM;
calculating, by the RLMM, a reward value based on the reward function during the operation of the interim version of the MLM;
determining by an operation comprising determining a provisional fairness value for the provisional version of the MLM based on the reward value;
determining that the interim fairness value is greater than the initial fairness value;
responsive to the determination that the interim fairness value is greater than the initial fairness value, replacing the initial version of the MLM with the interim version of the MLM, and replacing the initial fairness value with the interim fairness value. A computer-implemented method, comprising: replacing. 前記初期公平性値が所定の閾値を超えるまで、前記動作を反復的に繰り返すステップをさらに含む、請求項１に記載のコンピュータ実施方法。 2. The computer-implemented method of claim 1, further comprising iteratively repeating said operation until said initial fairness value exceeds a predetermined threshold. 前記初期ＭＬＭは教師ありＭＬＭである、請求項１または２に記載のコンピュータ実施方法。 3. The computer-implemented method of claim 1 or 2, wherein the initial MLM is a supervised MLM. 前記公平性関連目標値は、性別、年齢、国籍、宗教的信念、民族性および指向のうちの少なくとも１つを含む、請求項１ないし３のいずれか一項に記載のコンピュータ実施方法。 4. The computer-implemented method of any one of claims 1-3, wherein the fairness-related target values include at least one of gender, age, nationality, religious beliefs, ethnicity and orientation. 構成および読み出しに基づいて前記初期ＭＬＭを前記強化学習メタ・モデルにリンクさせるステップをさらに含む、請求項１ないし４のいずれか一項に記載のコンピュータ実施方法。 5. The computer-implemented method of any one of claims 1-4, further comprising linking the initial MLM to the reinforcement learning meta-model based on configuration and retrieval. 前記複数のパラメータ値は、以下のパラメータ・タイプ、すなわち、重み付け係数および活性化関数変数のうちの少なくとも１つの値を含む、請求項１ないし５のいずれか一項に記載のコンピュータ実施方法。 6. The computer-implemented method of any one of claims 1-5, wherein the plurality of parameter values comprises values of at least one of the following parameter types: weighting factors and activation function variables. 前記複数のハイパーパラメータ値は、以下のハイパーパラメータ・タイプ、すなわち、活性化関数のタイプ、層あたりのノードの数、ニューラル・ネットワークの層の数および機械学習モデルのうちの少なくとも１つの値を含む、請求項１ないし６のいずれか一項に記載のコンピュータ実施方法。 The plurality of hyperparameter values includes at least one value of the following hyperparameter types: activation function type, number of nodes per layer, number of layers of neural network and machine learning model. 7. The computer-implemented method of any one of claims 1-6. 請求項１ないし７のいずれか一項に記載のコンピュータ実施方法における各ステップをコンピュータに実行させる、コンピュータ・プログラム。A computer program product that causes a computer to perform the steps of the computer-implemented method of any one of claims 1-7. 請求項８に記載のコンピュータ・プログラムを記録したコンピュータ可読記憶媒体。9. A computer-readable storage medium recording the computer program according to claim 8. コンピュータ・システムであって、
１つまたは複数のコンピュータ・プロセッサと、
１つまたは複数のコンピュータ可読記憶媒体と、
前記１つまたは複数のコンピュータ可読記憶媒体に記憶されているプログラム命令とを備え、前記プログラム命令は、請求項１ないし７のいずれか一項に記載のコンピュータ実施方法における各ステップを前記１つまたは複数のコンピュータ・プロセッサに実行させるよう構成されている、コンピュータ・システム。 A computer system,
one or more computer processors;
one or more computer readable storage media;
and program instructions stored on said one or more computer-readable storage media, said program instructions for executing each step of the computer-implemented method of any one of claims 1-7 in said one. or a computer system configured to run on multiple computer processors.
JP2021558964A 2019-04-08 2020-03-18 Improving fairness through reinforcement learning Pending JP2022527536A (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US16/377,727 2019-04-08
US16/377,727 US20200320428A1 (en) 2019-04-08 2019-04-08 Fairness improvement through reinforcement learning
PCT/IB2020/052465 WO2020208444A1 (en) 2019-04-08 2020-03-18 Fairness improvement through reinforcement learning

Publications (2)

Publication Number Publication Date
JP2022527536A JP2022527536A (en) 2022-06-02
JPWO2020208444A5 true JPWO2020208444A5 (en) 2022-08-18

Family

ID=72663093

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2021558964A Pending JP2022527536A (en) 2019-04-08 2020-03-18 Improving fairness through reinforcement learning

Country Status (6)

Country Link
US (1) US20200320428A1 (en)
JP (1) JP2022527536A (en)
CN (1) CN113692594A (en)
DE (1) DE112020000537T5 (en)
GB (1) GB2597406A (en)
WO (1) WO2020208444A1 (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11984199B2 (en) * 2019-08-02 2024-05-14 Kpn Innovations, Llc Methods and systems for generating compatible substance instruction sets using artificial intelligence
US11636386B2 (en) * 2019-11-21 2023-04-25 International Business Machines Corporation Determining data representative of bias within a model
US11556826B2 (en) * 2020-03-20 2023-01-17 Adobe Inc. Generating hyper-parameters for machine learning models using modified Bayesian optimization based on accuracy and training efficiency
US11551178B2 (en) * 2020-05-14 2023-01-10 Wells Fargo Bank, N.A. Apparatuses and methods for regulation offending model prevention
CN112163677B (en) * 2020-10-14 2023-09-19 杭州海康威视数字技术股份有限公司 Method, device and equipment for applying machine learning model
CN112257848B (en) * 2020-10-22 2024-04-30 北京灵汐科技有限公司 Method for determining logic core layout, model training method, electronic device and medium
WO2022115402A1 (en) * 2020-11-27 2022-06-02 Amazon Technologies, Inc. Staged bias measurements in machine learning pipelines
CN112416602B (en) * 2020-12-10 2022-09-16 清华大学 Distributed data stream resource elastic expansion enhancing plug-in and enhancing method
CN112905465B (en) * 2021-02-09 2022-07-22 中国科学院软件研究所 Machine learning model black box fairness test method and system based on deep reinforcement learning
US20220391683A1 (en) * 2021-06-07 2022-12-08 International Business Machines Corporation Bias reduction during artifical intelligence module training
EP4106231A1 (en) * 2021-06-14 2022-12-21 Google LLC Selection of physics-specific model for determination of characteristics of radio frequency signal propagation
US20230351172A1 (en) * 2022-04-29 2023-11-02 Intuit Inc. Supervised machine learning method for matching unsupervised data

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9008840B1 (en) 2013-04-19 2015-04-14 Brain Corporation Apparatus and methods for reinforcement-guided supervised learning
US10839302B2 (en) 2015-11-24 2020-11-17 The Research Foundation For The State University Of New York Approximate value iteration with complex returns by bounding
US11176487B2 (en) * 2017-09-28 2021-11-16 Oracle International Corporation Gradient-based auto-tuning for machine learning and deep learning models
CN109242105B (en) * 2018-08-17 2024-03-15 第四范式（北京）技术有限公司 Code optimization method, device, equipment and medium

Similar Documents

Publication Publication Date Title
US11544630B2 (en) Automatic feature subset selection using feature ranking and scalable automatic search
GB2597406A (en) Fairness improvement through reinforcement learning
US11620568B2 (en) Using hyperparameter predictors to improve accuracy of automatic machine learning model selection
US11868854B2 (en) Using metamodeling for fast and accurate hyperparameter optimization of machine learning and deep learning models
US11481637B2 (en) Configuring computational elements for performing a training operation for a generative adversarial network
JPWO2020208444A5 (en)
US11615265B2 (en) Automatic feature subset selection based on meta-learning
US11468379B2 (en) Automated evaluation of project acceleration
US11562178B2 (en) Adaptive sampling for imbalance mitigation and dataset size reduction in machine learning
US20210374544A1 (en) Leveraging lagging gradients in machine-learning model training
Tran et al. Data-driven neural architecture learning for financial time-series forecasting
WO2022031561A1 (en) Memory usage prediction for machine learning and deep learning models
WO2020033559A1 (en) System and method for structure and attribute based graph partitioning
AU2021308941A1 (en) Account prediction using machine learning
JP2023162348A (en) Data sparsity monitoring during neural network training
Du et al. Inference of gene regulatory networks based on the Light Gradient Boosting Machine
JP7279225B2 (en) METHOD, INFORMATION PROCESSING DEVICE, AND PROGRAM FOR TRANSFER LEARNING WHILE SUPPRESSING CATASTIC FORGETTING
WO2022166125A1 (en) Recommendation system with adaptive weighted baysian personalized ranking loss
Djuric et al. Distributed confidence-weighted classification on mapreduce
US20240095231A1 (en) Expert-optimal correlation: contamination factor identification for unsupervised anomaly detection
US11921756B2 (en) Automated database operation classification using artificial intelligence techniques
KR102583120B1 (en) Apparatus and method for providing benchmark prediction result of artificial intelligence based model
US11429864B1 (en) System and method for bank-balanced sparse activation and joint-activation-weight-sparse training of neural networks
WO2024063828A1 (en) Expert-optimal correlation: contamination factor identification for unsupervised anomaly detection
US20230342672A1 (en) Method and system for classification and/or prediction on unbalanced datasets