Why increasing the number of units or layers does not increase the accuracy and decrease the loss?












1












$begingroup$


I have an LSTM neural network; when I increase the number of units, layers, epochs or add dropout, it seems it has no effect and still I have persistent errors and accuracies like the following:




loss: 3.5071 - acc: 0.0981 - val_loss: 6.7042 - val_acc: 0.0122




Why this happens and how can I fix it?






machine-learning neural-network deep-learning lstm loss-function







share|improve this question











$endgroup$












  • $begingroup$
    Those accuracies are so low that I wonder whether something is seriously wrong. Can you provide some more context?
    $endgroup$
    – Ben Reiniger
    yesterday
















1












$begingroup$


I have an LSTM neural network; when I increase the number of units, layers, epochs or add dropout, it seems it has no effect and still I have persistent errors and accuracies like the following:




loss: 3.5071 - acc: 0.0981 - val_loss: 6.7042 - val_acc: 0.0122




Why this happens and how can I fix it?






machine-learning neural-network deep-learning lstm loss-function







share|improve this question











$endgroup$












  • $begingroup$
    Those accuracies are so low that I wonder whether something is seriously wrong. Can you provide some more context?
    $endgroup$
    – Ben Reiniger
    yesterday














1












1








1





$begingroup$


I have an LSTM neural network; when I increase the number of units, layers, epochs or add dropout, it seems it has no effect and still I have persistent errors and accuracies like the following:




loss: 3.5071 - acc: 0.0981 - val_loss: 6.7042 - val_acc: 0.0122




Why this happens and how can I fix it?






machine-learning neural-network deep-learning lstm loss-function







share|improve this question











$endgroup$




I have an LSTM neural network; when I increase the number of units, layers, epochs or add dropout, it seems it has no effect and still I have persistent errors and accuracies like the following:




loss: 3.5071 - acc: 0.0981 - val_loss: 6.7042 - val_acc: 0.0122




Why this happens and how can I fix it?






machine-learning neural-network deep-learning lstm loss-function




machine-learning neural-network deep-learning lstm loss-function






share|improve this question















share|improve this question













share|improve this question




share|improve this question








edited 2 days ago









Media

7,35062161




7,35062161










asked 2 days ago









user145959user145959

1268




1268












  • $begingroup$
    Those accuracies are so low that I wonder whether something is seriously wrong. Can you provide some more context?
    $endgroup$
    – Ben Reiniger
    yesterday


















  • $begingroup$
    Those accuracies are so low that I wonder whether something is seriously wrong. Can you provide some more context?
    $endgroup$
    – Ben Reiniger
    yesterday
















$begingroup$
Those accuracies are so low that I wonder whether something is seriously wrong. Can you provide some more context?
$endgroup$
– Ben Reiniger
yesterday




$begingroup$
Those accuracies are so low that I wonder whether something is seriously wrong. Can you provide some more context?
$endgroup$
– Ben Reiniger
yesterday










1 Answer
1






active

oldest

votes


















2












$begingroup$

The title of your question and the question itself are somehow different but I'll try to answer the question, the meaning of decreasing loss without changes in accuracy.



The reason is simply due to using probabilities. For instance, for classification task if you have an output $0.7$ for input and the last layer is a softmax, then you classify the input as that class which has $0.7$. Imagine you train more and that output changes to something like $0.95$. Consequently, the accuracy does not change because you already classify it as what it really is but the loss lessens.



To answer the question which is in the body of your post, there can be numerous reasons that I'll try to refer to them.



One of the possibilities is that your data of different classes have overlap in the current feature space. This may lead to high Bayes error. For instance, suppose you have two same inputs and the label of them are contradictory. In this situation, your performance cannot be improved. To check it whether you've got this problem or not, take a look at the histogram of your data.



Another problem can be the weakness of LSTM networks which cannot memorise numerous things. LSTM models are very good at things like considering the gender of a subject or the plural or singular form of subjects but in cases where they should consider many things simultaneously, they have difficulties.



Another reason can be the incorrect way of using dropout. At first, do not use it and let your network overfits the training data to find a good model. After fitting your data, try to use dropout.



You can also test Stacked LSTMs which are powerful models.






share|improve this answer











$endgroup$













    Your Answer





    StackExchange.ifUsing("editor", function () {
    return StackExchange.using("mathjaxEditing", function () {
    StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
    StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
    });
    });
    }, "mathjax-editing");

    StackExchange.ready(function() {
    var channelOptions = {
    tags: "".split(" "),
    id: "557"
    };
    initTagRenderer("".split(" "), "".split(" "), channelOptions);

    StackExchange.using("externalEditor", function() {
    // Have to fire editor after snippets, if snippets enabled
    if (StackExchange.settings.snippets.snippetsEnabled) {
    StackExchange.using("snippets", function() {
    createEditor();
    });
    }
    else {
    createEditor();
    }
    });

    function createEditor() {
    StackExchange.prepareEditor({
    heartbeatType: 'answer',
    autoActivateHeartbeat: false,
    convertImagesToLinks: false,
    noModals: true,
    showLowRepImageUploadWarning: true,
    reputationToPostImages: null,
    bindNavPrevention: true,
    postfix: "",
    imageUploader: {
    brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
    contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
    allowUrls: true
    },
    onDemand: true,
    discardSelector: ".discard-answer"
    ,immediatelyShowMarkdownHelp:true
    });


    }
    });














    draft saved

    draft discarded


















    StackExchange.ready(
    function () {
    StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f47362%2fwhy-increasing-the-number-of-units-or-layers-does-not-increase-the-accuracy-and%23new-answer', 'question_page');
    }
    );

    Post as a guest















    Required, but never shown

























    1 Answer
    1






    active

    oldest

    votes








    1 Answer
    1






    active

    oldest

    votes









    active

    oldest

    votes






    active

    oldest

    votes









    2












    $begingroup$

    The title of your question and the question itself are somehow different but I'll try to answer the question, the meaning of decreasing loss without changes in accuracy.



    The reason is simply due to using probabilities. For instance, for classification task if you have an output $0.7$ for input and the last layer is a softmax, then you classify the input as that class which has $0.7$. Imagine you train more and that output changes to something like $0.95$. Consequently, the accuracy does not change because you already classify it as what it really is but the loss lessens.



    To answer the question which is in the body of your post, there can be numerous reasons that I'll try to refer to them.



    One of the possibilities is that your data of different classes have overlap in the current feature space. This may lead to high Bayes error. For instance, suppose you have two same inputs and the label of them are contradictory. In this situation, your performance cannot be improved. To check it whether you've got this problem or not, take a look at the histogram of your data.



    Another problem can be the weakness of LSTM networks which cannot memorise numerous things. LSTM models are very good at things like considering the gender of a subject or the plural or singular form of subjects but in cases where they should consider many things simultaneously, they have difficulties.



    Another reason can be the incorrect way of using dropout. At first, do not use it and let your network overfits the training data to find a good model. After fitting your data, try to use dropout.



    You can also test Stacked LSTMs which are powerful models.






    share|improve this answer











    $endgroup$


















      2












      $begingroup$

      The title of your question and the question itself are somehow different but I'll try to answer the question, the meaning of decreasing loss without changes in accuracy.



      The reason is simply due to using probabilities. For instance, for classification task if you have an output $0.7$ for input and the last layer is a softmax, then you classify the input as that class which has $0.7$. Imagine you train more and that output changes to something like $0.95$. Consequently, the accuracy does not change because you already classify it as what it really is but the loss lessens.



      To answer the question which is in the body of your post, there can be numerous reasons that I'll try to refer to them.



      One of the possibilities is that your data of different classes have overlap in the current feature space. This may lead to high Bayes error. For instance, suppose you have two same inputs and the label of them are contradictory. In this situation, your performance cannot be improved. To check it whether you've got this problem or not, take a look at the histogram of your data.



      Another problem can be the weakness of LSTM networks which cannot memorise numerous things. LSTM models are very good at things like considering the gender of a subject or the plural or singular form of subjects but in cases where they should consider many things simultaneously, they have difficulties.



      Another reason can be the incorrect way of using dropout. At first, do not use it and let your network overfits the training data to find a good model. After fitting your data, try to use dropout.



      You can also test Stacked LSTMs which are powerful models.






      share|improve this answer











      $endgroup$
















        2












        2








        2





        $begingroup$

        The title of your question and the question itself are somehow different but I'll try to answer the question, the meaning of decreasing loss without changes in accuracy.



        The reason is simply due to using probabilities. For instance, for classification task if you have an output $0.7$ for input and the last layer is a softmax, then you classify the input as that class which has $0.7$. Imagine you train more and that output changes to something like $0.95$. Consequently, the accuracy does not change because you already classify it as what it really is but the loss lessens.



        To answer the question which is in the body of your post, there can be numerous reasons that I'll try to refer to them.



        One of the possibilities is that your data of different classes have overlap in the current feature space. This may lead to high Bayes error. For instance, suppose you have two same inputs and the label of them are contradictory. In this situation, your performance cannot be improved. To check it whether you've got this problem or not, take a look at the histogram of your data.



        Another problem can be the weakness of LSTM networks which cannot memorise numerous things. LSTM models are very good at things like considering the gender of a subject or the plural or singular form of subjects but in cases where they should consider many things simultaneously, they have difficulties.



        Another reason can be the incorrect way of using dropout. At first, do not use it and let your network overfits the training data to find a good model. After fitting your data, try to use dropout.



        You can also test Stacked LSTMs which are powerful models.






        share|improve this answer











        $endgroup$



        The title of your question and the question itself are somehow different but I'll try to answer the question, the meaning of decreasing loss without changes in accuracy.



        The reason is simply due to using probabilities. For instance, for classification task if you have an output $0.7$ for input and the last layer is a softmax, then you classify the input as that class which has $0.7$. Imagine you train more and that output changes to something like $0.95$. Consequently, the accuracy does not change because you already classify it as what it really is but the loss lessens.



        To answer the question which is in the body of your post, there can be numerous reasons that I'll try to refer to them.



        One of the possibilities is that your data of different classes have overlap in the current feature space. This may lead to high Bayes error. For instance, suppose you have two same inputs and the label of them are contradictory. In this situation, your performance cannot be improved. To check it whether you've got this problem or not, take a look at the histogram of your data.



        Another problem can be the weakness of LSTM networks which cannot memorise numerous things. LSTM models are very good at things like considering the gender of a subject or the plural or singular form of subjects but in cases where they should consider many things simultaneously, they have difficulties.



        Another reason can be the incorrect way of using dropout. At first, do not use it and let your network overfits the training data to find a good model. After fitting your data, try to use dropout.



        You can also test Stacked LSTMs which are powerful models.







        share|improve this answer














        share|improve this answer



        share|improve this answer








        edited 2 days ago

























        answered 2 days ago









        MediaMedia

        7,35062161




        7,35062161






























            draft saved

            draft discarded




















































            Thanks for contributing an answer to Data Science Stack Exchange!


            • Please be sure to answer the question. Provide details and share your research!

            But avoid …



            • Asking for help, clarification, or responding to other answers.

            • Making statements based on opinion; back them up with references or personal experience.


            Use MathJax to format equations. MathJax reference.


            To learn more, see our tips on writing great answers.




            draft saved


            draft discarded














            StackExchange.ready(
            function () {
            StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f47362%2fwhy-increasing-the-number-of-units-or-layers-does-not-increase-the-accuracy-and%23new-answer', 'question_page');
            }
            );

            Post as a guest















            Required, but never shown





















































            Required, but never shown














            Required, but never shown












            Required, but never shown







            Required, but never shown

































            Required, but never shown














            Required, but never shown












            Required, but never shown







            Required, but never shown







            -

            Popular posts from this blog

            How to label and detect the document text images

            Image
            1 $begingroup$ This is what I mean as document text image: I want to label the texts in image as separate blocks and my model should detect these labels as classes. NOTE: This is how the end result should be like: The labels like Block 1, Block 2, Block 3,.. should be Logo, Title, Date,.. Others, etc. Work done: First approach : I tried to implement this method via Object Detection, it didn't work. It didn't even detect any text. Second approach : Then I tried it using PixelLink. As this model is build for scene text detection, it detected each and every text in the image. But this method can detect multiple lines of text if the threshold values are increased. But I have no idea how do I add labels to the text blocks. PIXEL_CLS_WEIGHT_all_ones = 'PIXEL_CLS_WEIGHT_all_ones' PIXEL_C...
            Read more

            Vallis Paradisi

            Image
            Coordinata: .mw-parser-output .geo-default,.mw-parser-output .geo-dms,.mw-parser-output .geo-dec{display:inline}.mw-parser-output .geo-nondefault,.mw-parser-output .geo-multi-punct{display:none}.mw-parser-output .longitude,.mw-parser-output .latitude{white-space:nowrap} 33°03′S 71°37′W  /  33.05°S 71.617°W  / -33.05; -71.617 Vallis Paradisi Res apud Vicidata repertae: Civitas : Chilia Numerus incolarum : 296 655 Zona horaria : UTC-4 Situs interretialis Nomen officiale : Valparaíso Gubernium Consilium : municipal council of Valparaíso Geographia Superficies : 401 600 000 Metrum quadratum Coniunctiones urbium Urbes gemellae : Ovetum , Badalona , Santa Fe , Ora longa , Sciamhaevum , Quantunia , Callao , Bat Yam , Melaka , Novorossiysk , Busan , Corduba , Odessa , Rosario , Malacca , Barcino , Soteropolis , Franciscopolis (California) , Guanajuato, Silao Tabula aut despectus Despectus in portum sinumque et urbem Vallis Paradisi Vallis Para...
            Read more

            Tabula Rosettana

            Image
            Tabula multilinguis Rosettana in Museo Britannico ostenditur. Tabula Rosettana, [1] etiam titulo OGIS 90 agnita, est stela decreto de rebus sacris in Aegypto anno 196 a.C.n. lato inscripta. Tabula iuxta Rosettam Aegypti, urbem in delta Nili et ad oram maris Mediterranei iacentem, anno 1799 a milite Francico reperta est. Inventio stelae, linguis duabus scripturisque tribus inscriptae, eruditis Instituti Aegypti statim nuntiata est; ibi enim iussu imperatoris Napoleonis eruditi omnium scientiarum (sub aegide Commissionis Scientiarum et Artium) properaverant cum expeditione Francica. Qua a Britannis mox debellata, tabula Rosettana Londinium missa hodie apud Museum Britannicum iacet. Textus Graecus cito lectus interpretationi textuum Aegyptiorum (in formis hieroglyphica et demotica expressorum) gradatim adiuvit. Denique textum plene interpretatus est Ioannes Franciscus Champollion. Ab opere eruditorum cumulativo coepit hodiernus scripturae hieroglyphicae linguaeque Aegyptiae a...
            Read more