Пример #1
0
def train_net(config):

    # UNPACK CONFIGS
    (train_filenames, img_mean) = unpack_configs(config)

    import theano.sandbox.cuda
    theano.sandbox.cuda.use(config['gpu'])

    import theano
    theano.config.on_unused_input = 'warn'
    import theano.tensor as T

    from multilabel_layers import DropoutLayer
    from multilabel_net import CNN_model, compile_models

    import theano.misc.pycuda_init
    import theano.misc.pycuda_utils

    # load hash_step1_bits
    group_idx = sio.loadmat('./step1/temp/group_idx.mat')
    group_idx = group_idx['group_idx']
    group_idx = group_idx[0][0]
    code_per_group = 8

    bits_idxes = range((group_idx - 1) * code_per_group)

    config['output_num'] = len(bits_idxes)

    model = CNN_model(config)

    batch_size = model.batch_size
    layers = model.layers
    weight_types = model.weight_types
    params = model.params

    val_filenames = train_filenames[:20]

    n_train_batches = len(train_filenames)
    minibatch_range = range(n_train_batches)

    ## COMPILE FUNCTIONS ##
    (train_model, validate_model, predict_model, train_error, learning_rate,
     shared_x, shared_y, vels) = compile_models(model, config)

    train_labels = None

    for idx in bits_idxes:

        hash_step1_code = h5py.File('./step1/temp/hash_step1_code_' +
                                    str(idx + 1) + '.mat')
        temp = np.transpose(np.asarray(
            hash_step1_code['hash_step1_code'])).astype('int64')

        if train_labels is None:
            train_labels = temp
        else:
            train_labels = np.hstack([train_labels, temp])

    train_labels[train_labels == -1] = 0

    val_labels = train_labels[:20 * batch_size]

    ######################### TRAIN MODEL ################################

    print '... training'

    #    initialize_weights(layers, weight_types)
    #    learning_rate.set_value(config['learning_rate'])

    #    vels = [theano.shared(param_i.get_value() * 0.)
    #            for param_i in params]

    # Start Training Loop
    epoch = 0
    step_idx = 0
    val_record = []
    predicted_labels = None
    while epoch < config['n_epochs']:
        epoch = epoch + 1

        if config['shuffle']:
            np.random.shuffle(minibatch_range)

        if config['finetune'] and epoch == 1 and not config['resume_train']:
            load_weights_finetune(layers, config['finetune_weights_dir'])

        count = 0
        for minibatch_index in minibatch_range:

            num_iter = (epoch - 1) * n_train_batches + count
            count = count + 1
            if count == 1:
                s = time.time()
            if count == 20:
                e = time.time()
                print "time per 20 iter:", (e - s)

            cost_ij = train_model_wrap(train_model, shared_x, shared_y,
                                       minibatch_index, minibatch_range,
                                       batch_size, train_labels,
                                       train_filenames, img_mean)

            if num_iter % config['print_freq'] == 0:
                print 'training @ iter = ', num_iter
                print 'training cost:', cost_ij
                if config['print_train_error']:
                    print 'training error rate:', train_error()

        ############### Test on Validation Set ##################

        DropoutLayer.SetDropoutOff()

        this_validation_error, this_validation_loss = get_val_error_loss(
            shared_x, shared_y, img_mean, val_filenames, val_labels,
            batch_size, validate_model)

        print('epoch %i: validation loss %f ' % (epoch, this_validation_loss))
        print('epoch %i: validation error %f %%' %
              (epoch, this_validation_error * 100.))
        val_record.append([this_validation_error, this_validation_loss])

        savepath = config['weights_dir'] + 'classifier_' + str(group_idx -
                                                               1) + '/'
        if not os.path.exists(savepath):
            os.mkdir(savepath)

        np.save(savepath + 'val_record.npy', val_record)

        DropoutLayer.SetDropoutOn()

        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx, val_record,
                                        learning_rate)

        # Save weights for each iteration
        if epoch % 5 == 0:

            save_weights(layers, savepath, epoch)
            np.save(savepath + 'lr_' + str(epoch) + '.npy',
                    learning_rate.get_value())
            save_momentums(vels, savepath, epoch)

    DropoutLayer.SetDropoutOff()
    # generate the labels
    for minibatch_index in range(n_train_batches):

        label = get_prediction_labels(predict_model, shared_x, minibatch_index,
                                      train_filenames, img_mean)
        if predicted_labels is None:
            predicted_labels = label[0]
        else:
            predicted_labels = np.vstack((predicted_labels, label[0]))

    hash_step2_code = {'hash_step2_code': predicted_labels}
    sio.savemat('./temp/hash_step2_code_' + str(group_idx - 1) + '.mat',
                hash_step2_code)

    DropoutLayer.SetDropoutOn()

    print('Optimization complete.')
Пример #2
0
def train_net(config):

    # UNPACK CONFIGS
    (flag_para_load, train_filenames, val_filenames, train_labels, val_labels,
     img_mean) = unpack_configs(config)

    # pycuda set up
    drv.init()
    dev = drv.Device(int(config['gpu'][-1]))
    ctx = dev.make_context()

    if flag_para_load:
        #  zmq set up
        sock = zmq.Context().socket(zmq.PAIR)
        sock.connect('tcp://*****:*****@ iter = ', num_iter
                print 'training cost:', cost_ij
                if config['print_train_error']:
                    print 'training error rate:', train_error()

            if flag_para_load and (count < len(minibatch_range)):
                load_send_queue.put('calc_finished')

        ############### Test on Validation Set ##################

        DropoutLayer.SetDropoutOff()

        this_validation_error, this_validation_loss = get_val_error_loss(
            rand_arr,
            shared_x,
            shared_y,
            val_filenames,
            val_labels,
            flag_para_load,
            img_mean,
            batch_size,
            validate_model,
            send_queue=load_send_queue,
            recv_queue=load_recv_queue)

        print('epoch %i: validation loss %f ' % (epoch, this_validation_loss))
        print('epoch %i: validation error %f %%' %
              (epoch, this_validation_error * 100.))
        val_record.append([this_validation_error, this_validation_loss])
        np.save(config['weights_dir'] + 'val_record.npy', val_record)

        DropoutLayer.SetDropoutOn()
        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx, val_record,
                                        learning_rate)

        # Save weights
        if epoch % config['snapshot_freq'] == 0:
            save_weights(layers, config['weights_dir'], epoch)
            np.save(config['weights_dir'] + 'lr_' + str(epoch) + '.npy',
                    learning_rate.get_value())
            save_momentums(vels, config['weights_dir'], epoch)

    print('Optimization complete.')
Пример #3
0
def train_net(config):

    # UNPACK CONFIGS
    (flag_para_load, train_filenames, val_filenames,
     train_labels, val_labels, img_mean) = unpack_configs(config)

    # pycuda set up
    drv.init()
    dev = drv.Device(int(config['gpu'][-1]))
    ctx = dev.make_context()
    
    if flag_para_load:
        #  zmq set up
        sock = zmq.Context().socket(zmq.PAIR)
        sock.connect('tcp://*****:*****@ iter = ', num_iter
                print 'training cost:', cost_ij
                if config['print_train_error']:
                    print 'training error rate:', train_error()

            if flag_para_load and (count < len(minibatch_range)):
                load_send_queue.put('calc_finished')

        ############### Test on Validation Set ##################

        DropoutLayer.SetDropoutOff()

        this_validation_error, this_validation_loss = get_val_error_loss(
            rand_arr, shared_x, shared_y,
            val_filenames, val_labels,
            flag_para_load, img_mean,
            batch_size, validate_model,
            send_queue=load_send_queue, recv_queue=load_recv_queue)


        print('epoch %i: validation loss %f ' %
              (epoch, this_validation_loss))
        print('epoch %i: validation error %f %%' %
              (epoch, this_validation_error * 100.))
        val_record.append([this_validation_error, this_validation_loss])
        np.save(config['weights_dir'] + 'val_record.npy', val_record)

        DropoutLayer.SetDropoutOn()
        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx,
                                        val_record, learning_rate)

        # Save weights
        if epoch % config['snapshot_freq'] == 0:
            save_weights(layers, config['weights_dir'], epoch)
            np.save(config['weights_dir'] + 'lr_' + str(epoch) + '.npy',
                       learning_rate.get_value())
            save_momentums(vels, config['weights_dir'], epoch)

    print('Optimization complete.')
Пример #4
0
def train_net(config):
    # UNPACK CONFIGS
    (flag_para_load, train_filenames, val_filenames,
     train_labels, val_labels, img_mean) = unpack_configs(config)
    if flag_para_load:
        #  zmq set up
        sock = zmq.Context().socket(zmq.PAIR)
        sock.connect('tcp://*****:*****@ iter = %i" % (num_iter))
                logger.info("training cost: %lf" % (cost_ij))
                if config['print_train_error']:
                    logger.info('training error rate: %lf' % train_error())

            if flag_para_load and (count < len(minibatch_range)):
                load_send_queue.put('calc_finished')

        ############### Test on Validation Set ##################

        #"""
        DropoutLayer.SetDropoutOff()

        result_list = get_val_error_loss(
            rand_arr, shared_x, shared_y,
            val_filenames, val_labels,
            flag_para_load, img_mean,
            batch_size, validate_model,
            send_queue=load_send_queue,
            recv_queue=load_recv_queue,
            flag_top_5=flag_top5)


        logger.info(('epoch %i: validation loss %f ' %
              (epoch, result_list[-1])))

        if flag_top5:
            logger.info(('epoch %i: validation error (top 1) %f %%, (top5) %f %%' %
                (epoch,  result_list[0] * 100., result_list[1] * 100.)))
        else:
            logger.info(('epoch %i: validation error %f %%' %
                (epoch, result_list[0] * 100.)))

        val_record.append(result_list)
        np.save(config['weights_dir'] + 'val_record.npy', val_record)

        DropoutLayer.SetDropoutOn()
        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx,
                                        val_record, learning_rate)

        # Save weights
        if epoch % config['snapshot_freq'] == 0:
            save_weights(layers, config['weights_dir'], epoch)
            np.save(config['weights_dir'] + 'lr_' + str(epoch) + '.npy',
                       learning_rate.get_value())
            save_momentums(vels, config['weights_dir'], epoch)
        #"""

    print('Optimization complete.')
Пример #5
0
def train_net(config, private_config):

    # UNPACK CONFIGS
    (flag_para_load, flag_datalayer, train_filenames, val_filenames,
     train_labels, val_labels, img_mean) = \
        unpack_configs(config, ext_data=private_config['ext_data'],
                       ext_label=private_config['ext_label'])

    gpu_send_queue = private_config['queue_gpu_send']
    gpu_recv_queue = private_config['queue_gpu_recv']

    # pycuda and zmq set up
    drv.init()
    dev = drv.Device(int(private_config['gpu'][-1]))
    ctx = dev.make_context()

    sock_gpu = zmq.Context().socket(zmq.PAIR)
    if private_config['flag_client']:
        sock_gpu.connect('tcp://*****:*****@ iter = ', num_iter
                    print 'training cost:', cost_ij

                if config['print_train_error']:
                    error_ij = train_error()

                    gpu_send_queue.put(error_ij)
                    that_error = gpu_recv_queue.get()
                    error_ij = (error_ij + that_error) / 2.

                    if private_config['flag_verbose']:
                        print 'training error rate:', error_ij

            if flag_para_load and (count < len(minibatch_range)):
                load_send_queue.put('calc_finished')

        ############### Test on Validation Set ##################

        DropoutLayer.SetDropoutOff()

        this_val_error, this_val_loss = get_val_error_loss(
            rand_arr,
            shared_x,
            shared_y,
            val_filenames,
            val_labels,
            flag_datalayer,
            flag_para_load,
            batch_size,
            validate_model,
            send_queue=load_send_queue,
            recv_queue=load_recv_queue)

        # report validation stats
        gpu_send_queue.put(this_val_error)
        that_val_error = gpu_recv_queue.get()
        this_val_error = (this_val_error + that_val_error) / 2.

        gpu_send_queue.put(this_val_loss)
        that_val_loss = gpu_recv_queue.get()
        this_val_loss = (this_val_loss + that_val_loss) / 2.

        if private_config['flag_verbose']:
            print('epoch %i: validation loss %f ' % (epoch, this_val_loss))
            print('epoch %i: validation error %f %%' %
                  (epoch, this_val_error * 100.))
        val_record.append([this_val_error, this_val_loss])

        if private_config['flag_save']:
            np.save(config['weights_dir'] + 'val_record.npy', val_record)

        DropoutLayer.SetDropoutOn()
        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx, val_record,
                                        learning_rate)

        # Save Weights, only one of them will do
        if private_config['flag_save']:
            if epoch % config['snapshot_freq'] == 0:
                save_weights(layers, config['weights_dir'], epoch)
                np.save(config['weights_dir'] + 'lr_' + str(epoch) + '.npy',
                        learning_rate.get_value())
                save_momentums(vels, config['weights_dir'], epoch)

    print('Optimization complete.')
Пример #6
0
def train_net(config):
    # UNPACK CONFIGS
    (flag_para_load, train_filenames, val_filenames,
     train_labels, val_labels, img_mean) = unpack_configs(config)
    if flag_para_load:
        #  zmq set up
        sock = zmq.Context().socket(zmq.PAIR)
        sock.connect('tcp://*****:*****@iter " + str(count)
            if count == 1:
                s = time.time()
            if count == 20:
                e = time.time()
                print "time per 20 iter:", (e - s)
                logger.info("time per 20 iter: %f" % (e - s)) 
            cost_ij = train_model_wrap(train_model, shared_x,
                                       shared_y, rand_arr, img_mean,
                                       count, minibatch_index,
                                       minibatch_range, batch_size,
                                       train_filenames, train_labels,
                                       flag_para_load,
                                       config['batch_crop_mirror'],
                                       send_queue=load_send_queue,
                                       recv_queue=load_recv_queue)

            if num_iter % config['print_freq'] == 0:
                #print 'training @ iter = ', num_iter
                #print 'training cost:', cost_ij
		logger.info("training @ iter = %i" % (num_iter)) 
		logger.info("training cost: %lf" % (cost_ij)) 
                if config['print_train_error']:
                    logger.info('training error rate: %lf' % train_error())
                    #print 'training error rate:', train_error()

            if flag_para_load and (count < len(minibatch_range)):
                load_send_queue.put('calc_finished')

        ############### Test on Validation Set ##################

        #"""
        DropoutLayer.SetDropoutOff()

        # result_list = [ this_validation_error, this_validation_error_top5, this_validation_loss ]
        # or
        # result_list = [ this_validation_error, this_validation_loss ]
        result_list = get_val_error_loss(
        #this_validation_error, this_validation_loss = get_val_error_loss(
            rand_arr, shared_x, shared_y,
            val_filenames, val_labels,
            flag_para_load, img_mean,
            batch_size, validate_model,
            send_queue=load_send_queue, 
            recv_queue=load_recv_queue,
            flag_top_5=flag_top5)


        logger.info(('epoch %i: validation loss %f ' %
              (epoch, result_list[-1])))
        #print('epoch %i: validation loss %f ' %
        #      (epoch, this_validation_loss))
        if flag_top5:
            logger.info(('epoch %i: validation error (top 1) %f %%, (top5) %f %%' %
                (epoch,  result_list[0] * 100., result_list[1] * 100.)))
        else:
            logger.info(('epoch %i: validation error %f %%' %
                (epoch, result_list[0] * 100.)))
        #print('epoch %i: validation error %f %%' %
        #      (epoch, this_validation_error * 100.))
        val_record.append(result_list)
        #val_record.append([this_validation_error, this_validation_loss])
        np.save(config['weights_dir'] + 'val_record.npy', val_record)

        DropoutLayer.SetDropoutOn()
        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx,
                                        val_record, learning_rate)

        # Save weights
        if epoch % config['snapshot_freq'] == 0:
            save_weights(layers, config['weights_dir'], epoch)
            np.save(config['weights_dir'] + 'lr_' + str(epoch) + '.npy',
                       learning_rate.get_value())
            save_momentums(vels, config['weights_dir'], epoch)
        #"""

    print('Optimization complete.')
Пример #7
0
def train_net(config, private_config):

    # UNPACK CONFIGS
    (flag_para_load, train_filenames, val_filenames,
     train_labels, val_labels, img_mean) = \
        unpack_configs(config, ext_data=private_config['ext_data'],
                       ext_label=private_config['ext_label'])


    gpu_send_queue = private_config['queue_gpu_send']
    gpu_recv_queue = private_config['queue_gpu_recv']

    # pycuda and zmq set up
    drv.init()
    dev = drv.Device(int(private_config['gpu'][-1]))
    ctx = dev.make_context()

    sock_gpu = zmq.Context().socket(zmq.PAIR)
    if private_config['flag_client']:
        sock_gpu.connect('tcp://*****:*****@ iter = ', num_iter
                    log_iter.write("%d\n" % num_iter)
                    log_iter.flush()
                    print 'training cost:', cost_ij
                    log_err_cost.write("%f\n" % cost_ij)
                    log_err_cost.flush()

                if config['print_train_error']:
                    error_ij = train_error()

                    gpu_send_queue.put(error_ij)
                    that_error = gpu_recv_queue.get()
                    error_ij = (error_ij + that_error) / 2.

                    if private_config['flag_verbose']:
                        print 'training error rate:', error_ij
                        log_err_rate.write("%f\n" % error_ij)
                        log_err_rate.flush()


            if flag_para_load and (count < len(minibatch_range)):
                load_send_queue.put('calc_finished')

            if count%20 == 0:
                e = time.time()
                print "time per 20 iter:", (e - s)
                
        ############### Test on Validation Set ##################

        DropoutLayer.SetDropoutOff()

        this_val_error, this_val_loss = get_val_error_loss(
            rand_arr, shared_x, shared_y,
            val_filenames, val_labels,
            flag_para_load, img_mean,
            batch_size, validate_model,
            send_queue=load_send_queue, recv_queue=load_recv_queue)

        # report validation stats
        gpu_send_queue.put(this_val_error)
        that_val_error = gpu_recv_queue.get()
        this_val_error = (this_val_error + that_val_error) / 2.

        gpu_send_queue.put(this_val_loss)
        that_val_loss = gpu_recv_queue.get()
        this_val_loss = (this_val_loss + that_val_loss) / 2.

        if private_config['flag_verbose']:
            print('epoch %i: validation loss %f ' %
                  (epoch, this_val_loss))
            print('epoch %i: validation error %f %%' %
                  (epoch, this_val_error * 100.))
        val_record.append([this_val_error, this_val_loss])

        if private_config['flag_save']:
            np.save(config['weights_dir'] + 'val_record.npy', val_record)
            np.savetxt(config['weights_dir'] + 'val_record_txt.txt', val_record)

        DropoutLayer.SetDropoutOn()
        ############################################

        # Adapt Learning Rate
        step_idx = adjust_learning_rate(config, epoch, step_idx,
                                        val_record, learning_rate)

        # Save Weights, only one of them will do
        if private_config['flag_save']:
            if epoch % config['snapshot_freq'] == 0:
                save_weights(layers, config['weights_dir'], epoch)
                np.save(config['weights_dir'] + 'lr_' + str(epoch) + '.npy',
                        learning_rate.get_value())
                save_momentums(vels, config['weights_dir'], epoch)

    print('Optimization complete.')